CFTC’s Technology Advisory Committee Public Meeting on February 14, 2018


>>Good morning, everybody. If
we could take our seats, please. Very excited to get started this
morning. It is great to see so much
energy and enthusiasm in the room today. It really matches
my excitement around these topics, and to finally
have our first TAC meeting after about two years, and only our third TAC meeting
in four years. You think the market ask technology has
changed since then, you’re right, which is why we have such a
robust agenda for the committee to
consider today. And it’s just — it’s wonderful
to welcome all of our members here, and I wanted to thank you
for being here and in some cases traveling across the country, over long distances to
be with us today. I know that you had to rearrange
your schedule on a short notice, given the cancellation of our
last meeting. But we’re really grateful to you for your
participation. And I would also like to welcome
some of our new members, or all of our new members for for being here. I
hope you know what you’ve gotten yourself into. We’re excited to
use all of your expertise to advise the
commission on your thoughts around FinTech and
your thoughts around market developments.
So let me just actually take a quick moment to express my regret that Columbia business school proffer
is professor is not able to join us today. And before I recognize the
capillary money and commissioner Ben ham, I want to thank Dan Gorfine, the
director of lab CFTC, interacting chair and
designated federal officer of the TAC for all of the work he
put into this meeting, as well as George,
whose energy and enthusiasm are contagious around these issues.
A committee, a meeting, and work flows are only as good as the
leadership and the staff behind them, and we have the best of
the best here at the agency. So thank you for all of your work.
With that, I would like to recognize chairmen Giancarlo. >>CHAIRMAN GIANCARLO: As the
TAC designated officer, he call this
to order. Before we get started there are a few items that I’ve
been asked to mention. Please ensure your microphone is turned
on so the webcast and teleconference audience can hear you. Change
the position of your place card. For TAC members participating by
phone, keep your phone on mute until you are ready to speak and identify
yourself before hand. And this may soup ironic —
sound ironic but refrain from using electronic items. .we want
to ensure full participation bial members of the TAC. I will turn it back to
Commissioner Quintenz who will introduce opening remarks.
>>COMMISSIONER QUINTENZ: Thank you, Dan. Yes, please, Mr.
Chairman. >>CHAIRMAN GIANCARLO: Thank
you, tech members, and to other
participates, and welcome to those watching here in the room
and on the webcast. Today well examine fin it can,
automated trading technology, cybersecurity, distributed
ledger technology, and of course virtual currencies. FinTech, and the enormous burst
of human creativity that underlies so much of it, is having a transformative
impact on trading, markets, and the entire
global financial system. She’s changes have far-ranging
implications for capital formation and risk transfer,
both here and abroad. Lost week I had the Hanker to
testify before the U.S. Senate banking committee, and I said
that the first element in the CTFC’s engagement with FinTech
and virtual currencies was to learn everything we can about
the emerging technology. Good public policy ask
regulation can only be built on a thorough understanding of the
subject matter and its potential application.
In May of last year, our agency was pleased to announce a launch of
its lab CFTC initiative. And soon after, the director and
chief innovation officer, Daniel Gorfine. In creating lab CFTC, we wanted
the agency to keep pace with technological innovation. So lab CFTC is the focal point
of enhancing innovation and fair
competition for the benefit of the American public. And it
helps to ensure that we can keep pace with changes in our markets and proactively identify
emerging regulatory opportunities, challenges, and
risks. Lab CFTC is situated within our office of general
counsel, which allows it to leverage the expertise of the
CFTC’s legal team and manage the interface between technological
innovation, regulatory morning, and existing
rules rules and regulations. Since its launch, lab CFTC has
held over 150 meetings with entities
raging from established financial firms to start-up
companies. Its conducted these meetings
through office hour incisions New York,
Chicago, Washington, and earlier this year, in Silicon Valley.
Lab CFTC also published its first primer last October on the topic of
virtual currencies, and soon it will seek to crowdsource topics
for innovation competitions. In selecting Dan Gorfine to serve
as the designated officer, the
commission sought to ensure that the efforts of the TAC and lab CFTC were mutually
reinforcing. Cross-pollenization and the
breaked down of agency silos helps modernize our engagement with rapidly changing
markets. Lab CFTC was formed with
bipartisan support. Whatever success it achieves
will be bipartisan success. In the end, we share the same
goal: To support our vitality national interest in maintaining the
world easiest deepest and vibrant risk
transfer markets in this digital 21st century. So this meeting
is timely. We see what is on the horizon. We must be prepared,
responsible, and well-informed. As we confront the challenges
ahead, we will continue to look to the work and ideas of advisory groups like
this one, the TAC, that is meeting today. Thank you once again.
>>COMMISSIONER QUINTENZ: Thank you, Mr. Chairman. Commissioner Behnam.
>>SPEAKER TWO: Thank you. I wish Commissioner Quintenz a
handicap Valentine’s day. [Laughter.] Thank you for his meeting,
scholarship sponsor IPship and leadership. You composed a comprehensive
agenda. I would also like to thank
Daniel Gorfine, director of lab CFTC,
and the TAC’s designated federal
officer, for planning today’s thoughtful and timely agenda.
Finally, I thank the panelists for sharing their time and
insight. It’s nice to see many familiar
faces. Committees like the TVAC and the
market risk committees, by providing a
directors diversity of views and a high level of expertise in matters
that carry systematic consequences.
These advisory committees ensure that the commission remains
abreast of market issues on the horizons that
could revolutionaryize markets
stability, transparency, or could shock
markets, increase risk and sulfate to fraud manipulation.
To today’s agenda and the year to come, I commend the committee for
endeavoring to tackle a wide range of topics and issues. Twice awarded Nobel Prize winner
Linus Pauling once said, the best way to have a good idea is
to have a lot of good ideas. With that in mind and the
combined knowledge in this room, I anticipate that the TAC is
going to have a great year. Leading off today’s meeting, are
virtual currencies and their underlying blockchain are
contributed ledger technologies, given the events of the last several months, which
included the historic listing of the big coin futures contracts, the price of Bitcoin
skyrocketing, and my personal education in Bitcoin, DLT, digital waltz,
and something called crypto kitty,
it’s no surprise that the TAC will be
exploring these development. As the CFTC continues to dive
into these topics, and ultimately shaping regulatory themes and oversight,
we just remain transparent and accountable to ensure that stakeholders and the
public may have an opportunity to contribute to these ask you
guys. Pivoting to the tutor — to these discussions. Pivoting to the future, as the
agenda suggestses, we need to encourage
dialogue between the commission and the stakeholders to ensure
that we provide a tailored oversight and customer
protections while ensuring accountableability.
Our regulations need to keep pace with technology, reflect
current technologies and demonstrate our capacity to
participate meaningfully in the adoption of new technology. In regard to automated trading technologies, the TAC plans to resuscitate some of regat. The proposes to establish
pretrade risk controls in an effort to mitigate the dangers
of an unchecked automated trading system. I’m happy to
see that this issue will be discussed today, because I think
it’s vitally important that the commission take immediate action
as an automated trading system that perhaps runs
amok can harm market participants. The market event,
flash, crash, or otherwise s not if, but when. I look forward to
the discussion today and to future discussions with industry participants regarding
thoughtful, appropriate regulation of automated trading.
In action, in regard to automated trading simply is not
an option. Lastly, I’m pleased that the meeting will close with
a discussion of cybersecurity, and emerging best trends and
practices. As I’ve said, each entity,
private Constitution must prioritize the safety of its
organization. It’s an all hands on deck exercise that demands
both financial and human resources to protect our
institutions, data, identity, and in many cases sovereignty. So that end, cybersecurity can’t
be views as territorial where we
seek to protect our own. Any attack on institutions
presents broad market risk and we must
view it as such to eliminate any symptoms.
I would like to acknowledge my support for the President’s FY2019
budget request. Its level setting at least in part the CFTC’s funding mechanism with
other US financial regulators. Since Dodd Frank was find law,
the CFTC and staff did a great job in meeting new
responsibilities and duties with limited resources. Combining
the new Dodd Frank responsibilities with existing
responsibilities and the continuous advent and
development of new technologies, including those discussed today,
mean that the CFTC’s overall duties will continue to grow.
I’m hopeful that these challenges will be recognized
and consequently supported with funding that at the at least
matches the President’s request. Too much is at stake and the
risk to the financial system are too great. I look forward to
hearing from the panels on all these important subjects, and
once again, I want to thanks Commissioner Quintenz, Daniel
Gorfine and the TAC. Thank you.>>COMMISSIONER QUINTENZ: Thank
you for your Valentine’s wishes. I hope
you enjoy your cupcake. Let me give my opening statement and we will turn it over to Dan for
the pal. Technology change has, in the derivatives markets, evolved
rapidly over the last 22 months since the last TAC meeting met.
I’m hopeful that we can build upon all of the robust work that
prior iterations of the technology advisory committee
have achieved. The committee today is going to explore five main areas, each of
which COMPLAEFRMifies how — the functionality of global direct evidence
markets. These are areas where the CFTC can demon freight
leadership and provide effective oversight that fosters the integrity, strength , and
liquidity of our markets. My intent is for the committee to
use this meeting to identify the
issues, and within each of these areas that it wishes to explore
in greater detail, with the ultimate goal of providing the
commission with actionable, practical advice.
In many cases, I anticipate that subcommittees may need to be
formed to enable the kind of focused review and thoughtful consideration to
arrive at those recommendations. The first area we will folk on
today is blockchain, and the potential application of DLT to
the derivatives market. DLT has the potential to transform how firms handle the execution,
processing, reporting, and recordkeeping of
derivative actions. Market participants my find that
use DLT to satisfy their observe
cautions results in efficiency and less cost but they may also
find that like with May many new opportunities in technologies, DLT also presents challenges.
Much work remains to be done in order to realize the full promise of
DLT, from scaleability issues, to DLT’s compatibility with existing CFTC
regulations that are many facets of DLT for the TAC to consider. Next, the committee will focus
on market and regulatory developments involvement virtual
because he knows related futures products. 87 the proliferation of virtual currencies, while
exciting from a innovation standpoint, raises questions and
challenges. Definitional questions about whether a
particularly viral currency or a token is a security on a commodity
continue to be debated. In addition to these foundational
legal questions, the growing demand to trade viral currencies also
elevate the risk to consumeers. I sure chairmen giancarlo’s view that we respect
the enthusiasm of investors for new digital currencies and meet
that enthusiasm with thoughtful, balanced regulation. The CFTC should not attempt to
make value judgments about which new
products are SCOMBORLT which are not. The markets, investors,
and consumers need to decide that for themselves.
However, the CFTC should aggressive target fraudulent and
manipulative behavior, before in the —
whether in the derivatives market or. I commend their efforts to protect the
regulatetory markets. As the markets for virtual
currencies mature, the commission, along
with its fellow state, federal, and international regulators,
should ensure a rational approach to regulatory
oversight, not one based on fear or inexperience.
Jurisdictional gaps should be identified and addressed, in
conjunction with those efforts. As I said last week’s yahoo! Finance summit, some SCO for cryptocurrency exchanges could
spur the development of spends around Caucasian security policies, data,
retention, protection of customer accounts, trading
practices and other issues. Self-regulation has a long
history in derivatives markets. It’s worst exploring whether an
SRO model could assist in
establishing standards. I look forward to hearing from the
committee about the possibility of such an SCO, and these are only a few
of the novel issues that the commission will gram with as this best of your
knowledgions asset class continues to evolve. I hope the
committee’s expertise can as assist the commission in
developing the opportunities presented by virtual because he
knows related financial products.
Next, the committee will dress machine learning, artificial
intelligence, and computing power. We’re going to have the
pleasure of hearing from Mr. Tim Estes, digital reasoning,
that developed a product that uses machine learning to
facilitate electronic communication surveillance. The
product uses machine learning to determine the meaning of words,
in emails and chats, kind of their connection, and the
program is also — context, and it’s designed to
flag potentially problematic behavior and mark those
communications for further review. He will speak with us
today about advances in machine learning technology and the
future impact on financial markets. Fourth, the committee will
examine the developments and challenges pose
bid the modern trading environment and automated
trading technologies. In the past, I’ve expressed my view
that proposed regulation, automated trading, or RegAT was a missed
opportunity by the commission to explore the real risks of the
modern trading environment. I believe the commission should
only pursue additional regulation in this space after it has identified
specific risks associated with automated trading, examined how those risks are
being addressed through the market incentives structure, and
then determine if regulation can play a proper role in
leafletting those — alleviating those risks. Once that threshold
determination has been made, the commission must work they feel to calibrate the scope of
the regulation with the specific risk in mind it’s meant to
target. The commission should not adopt automated trading regulations to
address amorphous, hypothetically
concerns, or simply for the sake of having them on the book. The TAC can serve as a forum to
reconsider the risks with a fresh eye. To the extent that
automated trading can be addressed through regulation,
the committee can look to explore what form that
regulation may take. I’m hopeful these kind of
conversations can better inform us as to the true risks posed by automated trading
and how we can best respond. Lisaly, the committee will
discuss cybersecurity developments. I expect this is
an ongoing dialogue between the industry and the
CFTC about what constitutes best practices. The TAC can
facilitate discussions about how the commission and market participants alike can improve
data transformation, storage,
archiving, and disposal processes. Thank you to all of our members, presenters, for being here to
take time to share your vision with us and rescheduling on such
a short notice. I’m excited to hear from you. Let’s get started.
>>Thank you all for your opening remarks. I would like
to build so what we just starred from Commissioner Quintenz and
discuss the scope and plan and approach. Given the emerging,
development, and impact of financial technologies across
markets, now is the time to be forward-looking and proactive —
and proif I have. We all need to consider how we can act to
ensure continued market enhancing innovation, market
integrity, and the leadership of our markets. To this end, the 2018TAC will be
clean focused on buffering work streams that can drive
actionable feedback and recommendations to the
commission. Today’s meeting will be broad
and seek to highlight key issues and topics involving financial
technology that merit further exploration by the TAC and
potential subcommittees throughout the course of the
year. Our format will be to begin with
brief presentations by our panelists, approximately 10
minutes each, followed by open discussion and questions from
our TAC members. At the end of our key subjective sections, we will — subjective
section, we will consider whether to create subcommittees
for further work will be appropriate. Before we get
started with our first panel, I want to recognize the work of
our colleagues in making today possible. Many have
contributed, though again, I do want to highlight my colleague, George Her ata, who has been instrumental
in keeping this together and keeping me sane during
postponement and rescheduling. I want to point out Michelle and
Bianca who have also contributed significantly to today’s
meeting. With that, let’s jump to our
first panel discussion. Which will cover DRVBT ledger
college initiatives and their impact.
Our panelists are Jennifer Peve from DHCC, Charley cooper from R3,
and Dan Bucsa from CFTC. If you would take the panel position,
we’ll get started with Jennifer. And all can move over there now.
>>DANIEL GORFINE: I think panel position might be a new
term. But … [Laughter. ] >>JENNIFER PEVE: Thank you,
Dan. So I’m Jennifer Peve, managing
director of the business development and office of
FinTech strange at DTCC. I thank you for the opportunity to
present to this committee today on technology innovation that is
really captured the tension of our industry over the last
several years. Blockchain or distributed ledger technology.
The rapid pace of change that is facing our industry today, while
nothing new continues to accelerate.
Technology innovations such as blockchain or distributed
ledgers have great potential to bring about transformative
change in our industry. In keeping up with such rapid
technology innovation is both a challenge as well as an impair stave as it effects
so many aspects of financial products and markets. It forces financial
substitutions and infrastructure providers to really consider out
emerging technology may disrupt our business, reshape how work
is done, make better use of massive quantities of data,
drive cost savings and modernize our infrastructure.
In addition, there’s a need to understand how supervision
oversight compliance rules and laws interact during this time
of such rapid technological change. It is essential that we
work together across industry organizations, technology
providers, market participants, regulators and policy makers to continue the
collaboration and develop best practices for technology
innovation such as they deliver on their promise. DTCC has been a strong advocate
for continued collaboration. It commends — the hub for the
agencies engagement with the FinTech innovation community. Together TAC and lab CFTC
provide great avenues to furthering the public-private
partnership as we as an industry continue to advance the
technology and address the challenges surrounding it. As a premier post trade market
infrastructure for global financial services try, DTCC understands
the financial transactions,
mitigating risk, increasing transparency.
DTCC is industry-owned and governed and has a 40-year history of
leveraging technology to drive innovation and reduce risks and
costs for our clients. We continue to build on that
legacy by advancing the use of technology to enhance post trade
processing while ensuring our applications are consistent with
key policy objectives, a risk mitigation, market security,
certainty, reliability, and efficiency.
DTCC actively explores how transform active focals like
blockchain distributed, machine learning,
can all enhance the post trade process
by further mitigating risks. In January 2017, DTCC mowned it
would reform the plat warehouse, using
a combination of cloud and distributed ledger. The TIW’s repository which
provides life cycle services for
approximately 98% of credit derivative
transaction in the global marketplace. There are four
main drivers for adoptic the ledger as a technology for TIW. The first was TIW needed a
technology renovation. Second, the preexistence.
Credit molds. Third, it provided DTCC and the industry
an opportunity to operationalize the technology on a business
with appropriate meaningful scale, while also mitigating
risk and driving cost efficiency for market participates. And
lastly, we had industry commitment. The decision to
move forward was made in collaboration with client design
partners who saw long-term value in moving to the combined cloud
distributed ledger platform. The project is being led by IBM
with Axoni and R3 acting an an
advisor. We work with our client design partners, financial market
infrastructure, SUV as IHS market, and industry standard bodies such as
ISDA. Throughout the development and implementation
of this effort. The initial phase of the project
is focused on two key — two key objectives. Fist, to minimize
the change in impact on market participants, and second, to
ensure the quality of the solution such that we minimize
any potential for operational risk to the industry. Post goal, we will roll out
future fallses of the project which will focus on node
administration and adoption, Lucian client and other —
allowing clients and other participants the option to adopt
nodes and become more active participates of the network. As withallowing client and other
— allowing clients and other participants the option to adopt
nodes and become more active participates of the network. As
with any technology innovation, operationalizeing the technology
helps to strengthen the solutions and can often
highlight limitations of the technology that must be dressed.
Our work on the TIW project over the last year has given us a
great deal of real-world experience in developing a replacement for a critical
industry-wide mainframe application. We’ve lettered what works —
learned what works and what doesn’t. We’ve seen the
technology’s limitations and discussed tradeoffs with our
technology providers and our design partner.
And we understand that this is still very much a work in
progress. One of the most important
aspects of deploying any distributed ledger use case is
in understanding the distributed ledger architecture. There are
a variety, and it’s important to perform due
diligence on each and examine how the approaches meet the
needs of a specific use case, as well as any potential tradeoffs.
Three examples of components of distributed ledger architecture
that one should examine include date
private and sharing. How is it broughted and for
whom? Data may only be distributed to
counterparents who are party to the trade but now consider how
validations on things like a trade identifier are accomplished when data is
propagated in many or — many disparate
distributed ledgers. There are also potential
implement indications for participants who initially elect nonnodal access
to a ledger network and over time decide to desktop a node.
How do you think about and migrate data from a nonnodal access
point to a full-on node? Data storage. Where is data soda snored on
chain, off chain and what implications are there for
meeting data retention policies that exist, whether you’re a network operator or a
participant on the network. Start contract. What level of
confidence and skills do you have and the language supported. How much functionality will be performedon chain versus
offchain and what if any lateens see is
introduced because of the use of the smart chain contracts and
where the functions are performed the? Other things, govern model.
Scaleability, and performance. DTCC appreciate’s the commission
as efforts to harmonize the best practices and encourages the
commission to continue this important work. In addition,
DTCC recommends that the commission consider the
following initiatives: Promoting common
global data standards and interoperability. A common set
of global standards could promote widespread
implementation and facilitate interoperable systems helping to
create silent systems. Second, recognize the importance of and
facilitate discussions concerning the establishment of governan
framework. The commission and policy makersers globally could
partner with industry leaders to provide guidance in the
establishment of such governance frameworks. I would be remiss if I didn’t
mention standards as another component. The development of common
standards and protocols required is critical
when implementing DLT. Distributed ledger technology
can enforce standards but it does not normalize or transform
the data. One of the initial use cases DTCC explored was
regulatory reporting. The global trade repository is the
only noncommercial global provider of transaction
reporting services for the CFTC and six other jurisdictions
across the globe. The GTR maintenance
approximately 40 million open derivative positions, and process over 1 billion messages
per month. Distributed legislator combined with other
technology have the potential to allow for more comprehensive transparency into the swap
market, provide enhanced over sites of the markets and a
deeper understanding of risk on a real-time basis.
Ultimately, a more efficient regulatory reporting framework. Advancing the application of
distributed ledger to support regulatory reporting is dependent on the consistency
of data elements and regulatory harm
nonization of reporting requirements on a dominance and
international basis should remain a key focus for the
industry. DTCC supports the efforts of
projects such as common domain model, a
project around standard using the industry’s representation of
derivative trade details and related processes, as well as
critical data elements. It encourages all in the industry
to work toward data standards. In condition, innovation
requires learning and experimentation.
We remain optimistic about the future of the distributed ledger
or blockchain. We continuously and actively experiment with the
technology, ensuring that distributed ledger initiatives
align with and support our business goals and objectives and deliver
client value. We understand it’s a nascent technology but
it’s evolving rapidly, and the changes we have today will be
solved tomorrow. Significant industry commitment
is required to realize the
potential of distributed legislator as changing industry
models takes time. Much of the value in the cases being
expoured comes in changes the business processes, not in the
actual use of the technology itself. And distributed ledger
technology success, defined by implementation, timing, participants’ ability to
adopt, business value it bring will
vary great lie by region and use case. Thank you.
>>DANIEL GORFINE: Thank you. Mr. Cooper?
>>CHARLEY COOPER: This is on? Yes. Thanks very much. Commissioner Egypt Quintenz,
thank you for the leadership and inviting me to participate today, and Chicagoer
jean Carlo, it’s very important. I would be remiss if I didn’t
also make a comment to the staff of
the commission. I’m an alumni of this building and the amount
of work to pull it off and the broad er initiatives rests
heavily on the shoulders of a small, lean staff that does a
hell of a job. A shotout to those guys in the room. Am I’m I’m going to go off
script, but I was thinking — I was on the plane, and I had this
sort of — I read the agenda and I thought I’m going to fit right
in, and then I got off the plane and I thought, no, probably not.
I think there’s — I think there’s a more — um —
controversial, but there’s a part of a conversation we need
to have with the private sector and the regulatory and how we
can work together that I think is often not flesh out as much
as it could be. So if you will permanent me, I’m
going to go a little bit off-script. Weies and how we can work
together that I think is often not flesh out as much as it
could be. So if you will permanent me, I’m
going to go a little bit off-script.
We live in this interesting time where technological innovation
and rapid technological innovation is something that’s
assumed. And we have a detention span
problem. This is the day and age of Twitter where we have massive tax policy
in 140 characters and we don’t watch three-hour movies anymore but watch them on
the subway in to work. But we get to the point where
technology evolves at all levels as rapidly as your smartphone.
I got a new phone the other day. And I’m not kidding. 24 hours
later, in the morning, there was a notification on my phone that
I find to update my software because the software that was on
the phone that I bought at the store the day before was not the
latest version. And we tend to think that
technological will evolve that quickly, and blockchain is I
think suffering in some ways from this expectations game, that the
initial white paper Bitcoin that came
out in 2009, nine years seems like a long time and you think
that complying blockchain would have taken over the world.
Smartphone technology are certain types of retail
technologies don’t operate in the same way that ledger-scale
— large-scale, enterprise-grade overhauls of the way systems and
processes work in not just financial services, but in all
sorts of businesses that blockchain could be applied to don’t change that —
that quickly. The first email was sent in
1971. That was before I was born. I graduated college in the
mid-’90s, and I didn’t have an email account. That is an intranet account at
Georgetown and we got homework assignments that I didn’t do,
which is not a surprise. If you think about it in that
context, the internet that is conceptualize in the ’40s and ’50s began in the
’70 and is didn’t come mainstream until the ’90s. I’m
into the saying that blockchain will take that along — I’m not
saying that blockchain will take that long. But the point s when
we have these conversations and people wonder where we are in
the evolution, why is it taking so long and you’re doing all
this experimentation, when you step back, you realize things
are happening quickly. From a large scale perspective. And the and the potential
impacts of this technology in financial services, specifically
in derivatives, but broadly across all a host of asset
classes, is massive. This isn’t a software update on a
smartphone. This is a much bigger deal. This, frankly,
goes to the very way in which capital markets operate, and if
we are right, if companies like
DTCC and ourselves and some of the others around this table are
right, this could represent over the next five to ten years a wholesale change in the
way these markets operate and consequently the way in which
you all, the commission, do your jobs.
This is a big deal. So that leads to the next
question, which is, okay, well, recognizing this is a big deal
and recognizing change is happening at whatever pace is,
where are we now? What does that mean in 2018? 2018 is the
year it’s about to get real. And what I mean by that is,
commercial deployments at scale in the marketplace of
distributed ledger solutions by regulated financial institutions
across asset classes. And I can give plenty of
examples and I’m happy to talk about in some detail, but there
have been years of approves of concept and people are board
with experiments and tests and whatever and people don’t seem
to think that’s interesting. That served a purpose. Before
we launch a commercial product into the marketplace, that
market SXARPTS regulators use, you damn
well better make sure it works. And that’s what the proof is
concept is about. Now we’re at a point where real applications
are going to be put in the market, and actually live trades
are being done as in the past couple
of weeks in our plasm platforms and
others, but this is beginning. Nowtform platforms and others,
but this is beginning. Now we’re real, okay?
And this is going to come to an example of what we’re doing, and
this plea that I’m going to make the commission.
One of the things we’re working on at R3 is regulatory
reporting. Regulatory reporting is enormous. All
sorts of different — every type of financial institution,
whether they’re a retail bank, an investment bank, a global
transaction bank, merchant bank, you name it, has requires to
report to you or some other regulatory in their home
jurisdictions, what types of activities, and
they have different ways of aggregating data, and what
information needs to be given, it’s a headache. So it’s not
like we’re going to go from no regulatory reporting so
blockchain regulatory reporting overnight. It’s not like the
industry shifts like that. What we believe is going to
happen, we’re going to start seeing folks on specific asset
classes, progressive the model, actually — proving the
model. Instead of DTC going from no
blockchain to blockchain overnight, they’re focused on
the information — within that part of the organization, to
make sure that it works before they make this wholesale
transition. Right? We think that makes sense. There’s not
this light switch that goes and suddenly the world wakes up to
blockchain. But to successful do that from
when we’re sitting, we need to take you all on that journey
with us. And that’s why the work of the
TAC and the CFTC lab and the rest are so important to what
we’re doing. But there’s a level beyond that,
that — this plea will come in a second, to get you involved.
One example is a regulatory reporting application we’re
building in connections with RBBS, and the
of course, and the UK. We’re — the FCC, and the UK. We’re
scoping what the application will look like and what they are
level of involvement will be, but they were involved in a
proof of concept that we did at the end of last year, focused
just on the mortgage market in the UK. Like I said, a discrete set
before you extrapolate in a bigger way. But this was an
example of a federal government regulatory in a major country, not just sitting back.
They were in the lab with us, with technologists, and with
regulatory specialists and experts helping
you us to develop a proof of concept that could be turned
into a product and getting advice on that.
That’s a very specific example. And the agenda talk about
regulatory reporting about that, but this could be extrapolated
out in all sorts of different things that affect
regulators from KYC to risk management to you name it. The monetary authority of
Singapore is involved in so much I can’t keep track. Almost tirings the bank of —
two years ago the bank of Canada — that was in a test
environment, but just to give you an indication of where they
were, that was the summer of 2016. One of these things that
we feel is really, really critical is getting regulators
involved in at that journey with us. At the end of the day, the
worst mistake we could make as private
sector entities is to build something we think is cool and fantastic and robust
and we get ready to launch, and we then
tell you and you say nice try but it doesn’t work for the
different policy reasons. Then we’ve wasted millions of
dollars in development, time and energy, only to find out that
what we want to provide to the masterpiece runs afoul of the
law in which you guys operate. We want to pull you guys in.
And so this is the plea. Put bluntly, we would ask, as passionately as possible, for
the US regulators and the governments,
to become even more active than you already are. The CFTC lab I think is one of
the only — but it’s certainly the most well-known of all the
regulatory efforts in this space. You’re leading the way
compared to the other regulators in Washington. But I can tell
you, there are federal governments around the world,
industrialized nations that are way outpacing the US government.
And that’s a concern. Because fully a third of our
clients, partners, investors, whatever you call them, we lead over 300 financial
and technology companies are US based. They report to you, and
they report to the Fed, and the SEC and have
different regulatory regimes they need to
take into account. Their competitors in Europe,
Asia, other places are gaining an
edge. And we would suggest, we’re
global. Right? We’re headquartered here but we
have more people abroad than we do in the US right now. We’re
involved in a level of sophistication and actual active
involvement from different regulators around the world to a
level we don’t see here in the US yet. So the plea is to the extent
that we can up that engagement and actually
— chairman, a leader in the space and a number of different
comments to get in there guys, like we need to take this
seriously, we need to procurement innovation, we could
not — promote innovation, we could not agree with you more. If we don’t do it right, we’re
going to mess ourselves up and what you
all are trying to do, and mess up your mission. This
technology is not just for the private sector. If this
technology is built and deployed appropriate, it will make your
jobs easier, make you more effective regulators.
An active participation — I’m not talking about just writing a
bunch of regulations before we know what
the technology is capable. I’m talking about technological and
business involvement, where we bring you all into this June with us
journey with us so you really begin — to understand the
technology does help us to develop it in an appropriate way
that meets your needs as well as the private sector needs. I’m
going to leave it a that. I know we’ll have a ton of
questions or whatever. I no, he that probably didn’t answer the
topic on the agenda but this is a forum that it just — I wanted
to use it to sort of throw myself at the
altar and beg for whatever I can. There you go. Thank you
very much. >>DAN BUCSA: Thank you. Mr. Bucsa. >>DAN BUCSA: Charley, here’s
my card. Looks like we’ll be talking. And I now get to follow up that impressive, off the cuff remarks
with a speech from a regulator. I’m a deputy director within the subdivision of market oversight. This is to represent a
regulatory use in reporting. For the disclaimers, the views
I’m about to share are mine and mine alone. They don’t
represent the commission or any of the staff.
Pleased to see familiar faces. I welcome the new members of
TAC, but more important than valuen tune’s day, it’s when pitchers and camps
catchers report for baseball, especially
the Cubs. Consider the careers of Rafael
Palmero as we juxtapose with blockchain
reporting. Feel free to Google the players.
Similar to when these players were Cubs, DLT is nascent and
carries potential and uncertainty. How the Cubs react
to the risks and the with regards of the two players as a remaineders of how not to
approach DLT for reporting. We need to take the time and make
an effort to evaluate costs and
benefits of DLT, whether it makes reporting better cheaper
and easier from.^ my perfect, the benefit of the TAC is to be
forward looking and have industry advise the industry.
By forward-looking, I mean years into the future and not the next
news cycle, election, or market shock. The chairman has
expressed a desire to transform away from being an analogy
regulator, and that’s not expected to happen overnight.
This is not the place for incremental steps, but we
encourage you to think big. I don’t want to focus on slight
improvements, but whether we reshape the reporting regimen via 5 to 10
year strategic to enact a significant win that helps market participants,
regulators, and those who depend on it. The future of oversight must
incourt DLT as it continues to improve and mature. The commission must stay abreast
of technology innovation and
protect the markets. The participants hope in taking
future technology to keep us informed, work in lock step with
regulators and consider regulatory needs during and not
after the development of applications.
Trying to adapt a system to meet regulations as an afterthought
is often costly and inadequate. The CFTC having a Saturday at
the table affords you say the opportunity to provide input and minimize the
burden of regulatory requirements. They must remain
technology neuter as much as possible, requires formats comply with principles without
prescribing how they’re met. As the requires will apply to
market participates regardless of the technology they utilize, if rules need to be
reevaluated for novel technologies.
Contemplating this interaction regarding blockchain between
regulator agencies and those they over see prompts questions.
Could DLT benefit reporting parties and the regulatory
community? Would DLT provide — what would
the future state of reporting look like? Can the TAC help us
determine if our rules are permissive of new technologies,
in not where friction points exist?
Despite this list of questions, we’re encouraged, the regular
tore use of technology could offer advantages. The first,
better regulation through technology.
DLT brings exciting potential for making reporting systems
more reliable, more automated, and less resource intensive.
The evolution of DLT could allow regulators to access data
automatically and seamlessly every time a trade is posted on
a particular blockchain without the need for human intervention
or interMedes immediateries. It could increase the speed and
improve the reliability of data. Commission access could be
incorporated into distributed ledgers of reporting parties.
This would allow for the commission tore more nimble and
efficient as it’s updated as they happen, real-time oversight
of market. Second advantage. Sharing of
data and greater access via DLT. As the financial crisis
demonstrated, the derivatives market is interconnected and global, requiring regulatory
coordination. Access is a part of the process. The TAC would
explore how to apply FinTech to the sill silo
approach. CFTC could collaborate with other authorities on best practices to
support nodes, undistributed ledgers. Imagine a setup where DLT
transcend the fragmented structure by providing reference
to a single, validated record of all financial information.
Some of your likely required to provide data to various — regulators in
an uncoordinated fashion. Those regulators duplicate their
efforts to parse that overlapping information. From the regulators viewpoint,
jurisdictional turf wars over selection and access protact the data
sharing agreements, and transfers of
sensitive data would be remnants of a
bigone era. Agencies would no longer be privilegely on only a sliver —
privy to no longer a sliver. Instead, every US regulator,
whether tasked with manipulation,
monetary policy or risk, would have immediate access to all the
data and make fully informed decisions.
Third advantage, reporting via blockchain. DLT should bear fruit from the perspective of the practitioner.
Market participants would be be a solved from having to create
new systems to maintain and transmit records, trans-pose data to match
regulatory formalities, complete a list of forms, and connect you and sometimes
pay for intermediaries to report. While the commission is
disincleaned to management the processing, the aim is to maintain fertile ground.
Since the agency would not force the use of DLT, the CFTC would
leverage the transition of the new technology if, and when,
market participants on their own accord decide to transition to a
new way of doing things. Blockchain will most likely only
be adopteded if individual firms decide it decreases burdens and increases
return on investment. Staff will not be recommending that
the commission make that investment choice for firms. Ensuring
issues that need to be involved are not meant to dissuade but highlight up knowns while
tempering enthusiasm the regulatory DLT is
both a definite type of. The use of DLT will need to be
subject to uniform standards that allow staff to successful
access and analyze the data. The current perception is that
data itself and the methods of its transformation are not
sufficiently standardized. The same issues that plague the
inception of any reporting stream would manifest themselves
with inconsistency of format if it’s not
coordinated. Interoperability becomes a
certain for regulators. Without consistency and if too many types of derivative ledgers
persistent, we could be saddled with
attempting to accommodate different ledgers.
— would be nearly unusable by the commission.
In parallel with how the agency would not compile participants to use
FinTech, the CFTC would not be likely to build interfaces for
DLT and traditional data in our business model with our
budgetary straightens. Second unknown.
CFTC retasking itself to use DLT. Even if interoperability
approaches, such as the ISDA common domain model, would the CFTC have the with
where about all to understand and consume information on the
blockchain? The CFTC has expended time, money, and
intellectual capital to build tools and delivery
methods. Before buffering DLT, the TAC
would need to develop a long-term technology plan. This
must include a determination of if approximate and how to
allocate fund to rebuild infrastructure and analyst
methods to take advantage of FinTech as the mechanism of the future.
Overpromising and underperforming could be risky if not planned.
Another challenge presents itself if DLT cannot make itself
trifecta to all entities — attractive to all
entities. It’s conceivable that large, sophisticated, and
technologically enhanced players will embrace
FinTech and blockchain’s become the linchpin of some parts of
financial activity. At the same time, it’s equally plausible
that some entities with reporting obligations reach the
opposite conclusion and decide not to abandon their existing
systems and spend time and money for DLT. But it appears harder to figure
out the details, and implement the plan by answering the
questions raised today. The take away is that we should try
to explore the technology and fully evaluate its long-term
merits. We should not dismiss the role it could play in the
regulatory space since it is new and unproven,
similarly young Jamie and Rafael. So why do I bother with you
reference to ball players in the 1980s the cubs lacked the vision
to appreciate what they could eventually come patience quickly
ran out as doubt overshadowed confidence and shift in
organization in the same trade. More rel had a fast ball 80
miners, and shoulder trouble. Would never hit for power to
justify starting role. When it was said in done more rel was a starting player for a twhopg
25 years an all star and series champion. Paul mer 0 was only 5 players in
the Blockchain reporting discussion,
we should not repeat the Cubs
mistakes and ignore the technology or shy away from
carving out a new path. Instead we should welcome the
challenge as growing pains can be over comd and nurtured
correctly not dismissing regulatory applications but
highlighting issues that need to be resolved at the same time we
must be cautious not to go all in without being cognizant of
risks. We need to tax expertise to
inform us if we are thinking of the right ecosystem for DLT and
reporting. Have we identified the appropriate challenges? How do we mitigate the concerns
if the commission were to proceed down this road? To sum
up if the feeling of adoption could benefit both market
participants and regulators, by decreasing burdens, increasing
standardization, and improving immediacy of oversight we should
take advantage of the current opportunity of this exciting
technology. Thank you.>>Thank you Dan I’d like to
thank our panelists and I will say as a
Red Sox fan I refer to think of Jamie
Moyer and Boston. I’d like to open it up to questions amongst
members and have in mind what are the potential work streams
that we would think could be the subject of subcommittees during
the course of the year, and then we’ll consider whether we would
want to recommend the creation of a subcommittee. I’ll throw
out a first question, you’re welcome to pick it up or ask
your But to pick up on what Dan had
mentioned, what are folks’ thoughts in terms of the current business
case, to don’t DLT or these types of ledger technologies?
Is there currently a business case, and what do we expect
adoption to look like? What’s the pace of adoption
going to be? And if you could do the name tag vertical on the — then we can
go that way. We’ll proceed that way.>>Chuck ochoa nex group. We’re a market provider of all
kinds of services, and I’m primarily
focused on DLT efforts surrounding post-trade services. The number one thing — I mean,
and talking to lots of other people in the marketplace, that we get from
distributed ledger technologies, are very different from sort of
the broader spectrum of what people had been interested in for blockchain
initially, you know we don’t need things like anonymity, we don’t necessarily
need truly trust as consensus, that’s not truly what you want.
The number — there are two things that we really get. We
get this notion of proof of existence, and
nonrepudiateability, which means if we blish a result out to the
world we can’t take it back — or we could prove to the world
we published it, people will know that it’s been published
and it’s never been tampered with, so, you know, nobody can
say you never published that, we can prove that it exists, and we
can never take it back That seems to be the two primary
things that people are really, really concerned about. A lot
of the services that we provide are, you know, value added
proprietary services that require, you know, tremendous
amounts of computation or tremendous amounts of external
data, so things like smart contracts on the blockchain
aren’t really feasible. You know, that’s where things like, you know, corda from R3 does
some of the thinking correctly. But those are the primary things
that people seem to really care about, they want to have this
one version of the truth or they want to have a source of one
version of the truth that everybody can, to be sure has ever been
tampered with or hasn’t been change in an
in audible way you can change a record as long as it’s a record
of the history. So those are the things that
people seem to be most concerned about and that’s sort of the
primary thing people seem to want to get out of this.>>
Daniel Gorfine: Thank you and I’ll>>Thank you. I wanted to
key off something that Mr. Cooper said about the regulators
coming along on the journey. And I think that not to take
away from that vision, because I think there’s a bigger there
that should be considered but I want to flag some of the
potential questions that we should be asking when we’re
talking about that.In particular, you know, if we’re
going F of the regulator serve as kind
of a consultant to the firms, how do we make certain that
that’s fair? We wouldn’t want a world where the regulator serves
as a consultant to R3, and none of R3’s competitors because that
would give R3 a significant advantage. We also need to I think be
thinking about how are we going to avoid
— put it this way how do we have a situation where we
wouldn’t want to have this be mandatory, right, like the only
way you can come to market is if you contract — you get a
consultancy with the regulator. Presumably we want a world where
people can still in good faith develop a product, and then
bring it forth, and see if it works or not face
whatever And finally, we should be
thinking about when we structure this, you
know, and no disrespect to our very gracious hosts here, but
like we don’t want a world where the regulator is like a vampire,
once you let them into their house you’re powerless to stop
them, right? We need to be thinking about how are we going
to use — how do the regulators use their data, or how — what’s
going to happen with that data, right?If the regulators have
access to data, how can it be used, under what terms, and how
is that structured. Which is all to say that I think
there’s certainly something to be said for the regulator being engaged early
and in a collaborative way, but we need to make certain that
it’s done well.>>Daniel Gorfine: Thank you,
and Ms.>>Great thank you very much to
the panelist for an insightful
presentation. My comments were getting a sense from you about
the integration of DLT technology into the workings of
the clearinghouse, and in particular looking at the sort
of many new risks the clearinghouses are facing
following cycle 7. And the advantages or
potentially the risks that are being created by DLT in relation to subsequent risks and
county part party risk in that regard t obviously the complication of
this will extend much wide ner that example for example in
relation to the requirements, how regulatory is going to
happen more broadly being I want to understand more precisely how
some of the latency issues in relation to the DLT data
reporting and robustness for example might translate into an impasse going
forward as the consequences of clearinghouse and mitigation in
that regard.>>So there are a lot of questions in there. I think — I’ll start with I
think as a clearinghouse of equities transactions, income and
equities, one of the challenges that we face when we look at
distributed ledger technology right now is the performance and
scalability aspects. So addressing and understanding
how many transactions you need to process in a second, what
kind of time to clear, and settle,
requirements you need to meet, and ensuring that whatever
platform you’re exploring, experimenting with, can support
that type of volume. Not only the volume that you
transact on a per second basis, per day basis, but also what
peak volumes you might need to hit in order to address any
regulatory requirements.So I think that’s one of our biggest challenges, I will tell you that
DTCC processes upwards of 25,000 transactions per second, in clearing
equities, and that’s something that we don’t believe currently
a distributed ledger can support. And that’s not just
instantiating the trades, but you have to consider putting the
trades on the blockchain, the actual process by which those
transactions are validated, what kind of processes are also being run on the
blockchain, and any latency involved, how many nodes can you
scale up to before you see, you know, any degradation in
performance.So there’s lots — and that’s not all the factors
in a go into it, but there’s lots of different — you know,
architecture components that really help And again, each ledger it’s
really important to look at multiple
different slavers of those ledgers because they all have
different capabilities then you can play with some of the levers
to understand what works or doesn’t work. In terms of risk a distributed
ledger can certainly assist with reducing settlement and
counterparty risk but it’s not the only solution, right? And again, I think for
clearinghouses, there is, you know, because of the performance
and scalability challenges, it is some time before they — you
know, are likely to be adopted in that capacity. And the reporting insight
obviously there’s a great use case for it, but it’s so
dependent upon the harmonization of the data standards across,
you know, on a global basis that until we
can address those data elements and
standardization around those data elements it makes it
difficult. Not impossible, you know, technology can solve
things every day, right? So there are certainly there’s
likely to be work-arounds but you have to evaluate our
work-arounds going to be worth the outcome versus, you know,
standardizing the data elements and getting a more streamlined
input and output.>>Daniel Gorfine: Mr.>>Thank
you, and again with everybody else I — this meeting I commend
the work of the commission and Stan in particular for
organizing — Dan in particular for organizing this.One
reaction, and one question. So Charley, when you talked about
why, you know you talked about the evolution of the distributed
ledger and the blockchain as being slow, I
agree with you. But the reality of life is you’ve got to step
back. The blockchains — I’m not sure
there is one blockchain out there, but the blockchains out
there were developed initially as decentralized distributed
ledgers. And so the rules of governance
are still evolving. The way to change any particular
blockchain requires a consensus. That isn’t necessarily the most
rapid, and maybe over time the process of governance will
improve, and maybe the methodology of consensus will —
you know, get a little bit faster and better, and things
will pick up. And I guess that leads me to
just an observation. I think that at least what I’m hearing
today, are really there’s two types of distributed ledger
technologies in play There’s the decentralized
distributed ledger technology, where nobody in particular necessarily runs it,
and it’s just a cryptography and encryption and mathematics that
hold it together and the rules of whatever the
assessment is, and then there’s the private or
permission distant distributed that relies on the technology. I wonder if Jennifer or Charley
or Dan have a view regarding the viability of both models. Is there a place for both
decentralized the distributed ledger technology and the
multiple of blockchains and the multiple of opportunities that
are out there, or is it only — are we only speaking about the
permission? And one of the reasons that I
mention that is because most of the applications I see, most of
the consortiums that I see today are the big folks, they’re folks
who are trying to take advantage for existing businesses.To the
extent you’re looking at some of the decentralized applications
that might be sitting on top of the
imperium chain maybe the innovators new guys trying to
figure out a way of doing something. Are there different approaches
is there different viability for both?>>Thank you. I’m going to
take two more questions or comments, Mr. Durkin then we’ll come over to
Mr. Stein.>>My question — first of all,
thank you for your presentations, you
all did an excellent job. And I want to just jump on a
theme that Jennifer in particular has
been underscoring, and I think we walked a similar journey together with
the trade repository services and the requirements that came, you know, via
legislation, as well as the requirements from the CFTC. And that is data standardization
and harmonization is extremely important. We at the CME group we think
blockchain has promise for the financial services sector, and we ourselves are
continuing to explore various applications for our We also feel it’s incumbent upon
all of us to pay attention to the work
and the opportunities associated with
joining various initiatives to help us understand and be part of
driving the standards, the policies,
associated with this technology. For example, we’re a
longstanding member of the Linux foundation, in
faction we’ve been actively engaged in developing their
hyperledger initiative, and part of that goal is committing ourselves to fostering open
standards, encouraging diverse
participation of market participants, cultivating the
highest standards and protocols as this So trying to be at this in the
formative stages I think is an imperative,
and an opportunity for the TAC to be focusing on this as part
of a possible subcommittee. We’re also part of the PTDL,
which is the post-trade distributive ledger group and
this is based out of London it’s made up of the largest global
banks, the exchanges, the clearinghouses the custodians
and global regulators, and we’ve been brought together, and we
were one of the core founding members
to understand how is this technology going to emerge, how
could it possibly be applied in terms of distributed ledger
technologies to our industries and our businesses that we
represent. We also continue to support
fixed protocol initiatives, and we serve on the fixed digital
currency and blockchain working group as the co chair and member
of that. So the point being made here, there are
opportunities, this is evolving, we do think that there’s
applications in various ways, and we commend the commission
for taking this topic on. And would ask that you consider it
as a possible subcommittee.>>Daniel Gorfine: Thank you.
Mr.>>So thank you. We laud the commission and this
group for looking at this exciting technology, clearly has
dramatic potential. However, I’d also like to point
out that technology on its own rarely solves a problem. And I
encourage this group, when you’re thinking about the
application of blockchain technology, we equally think In a proof of concept, in a
white room, you can do lots of interesting things. But we have a very
sophisticated, expensive financial infrastructure. And so an equal amount of
thought and work on how to get adoption is important. And also, I’ll say that those
like DTCC and the CFTC, the infrastructure, the regulators, are most likely to
be the source of the driving vision. There are members of the
ecosystem, the buy side, that are unlikely on their own to create a standard that
everyone will magically rally around. And so I encourage this group to embrace the full ecosystem, and
the buy side in particular, as part of these discussions to
ensure we both have a path to adoption, and one that takes into account the varied
perspectives of the different The last point I’d make to some
of the discussion is there’s a lot of hyperbole in blockchain and a lot of
skepticism as well. I think it behooves us to take a
practical approach to look at these innovative concepts that satoche
truly had some break throughs not in the underlying technology
but in how to combine none technologies. So look to figure out what is
the most practical solution, so that we
can achieve what Dan buck ca
described as the efficiencies and the clarity without undermining the ownership and
control of the intellectual property that I believe Bryan
mentioned. Thank you.>>Daniel Gorfine: Thank you all of our
panelists and thank you for the comments and questions posed.
And I think that to summarize this conversation, I think we’ve
just hit the tip of the iceberg. There’s much to explore in terms
of potential promise, as well as challenges, and real world
challenges. So I think based on what appear
to be a number of trends or themes or work streams worthy of
further exploration, I would, as the acting chair, move that the
TAC recommend to the commission that it consider creating a
subcommittee on Is there a second on that
motion? There are many seconds on that
motion. Are there any questions or comments, before we move to a
full vote? Okay, and if not I will now call
for the vote on the motion. All those in favor of recommending
to the commission that the commission consider creating a
subcommittee on distributed ledger technology, please say
aye. All of those opposed, please say
nay.Are there any abstentions? And we will now turn to our next
panel, and it was well set up by R by Mr. Stein talking about satochi and virtually currencies, our next
panel will discuss virtually
currencies, future products and market and regulatory
challenges. Our next Jeri Brito from Coin
Center, Gary DeWaal from Katten Muchin
and Rosen man, Richard Gorelick and Amir
zaidy from CFTC. So if the panel will take their positions
we will begin with Jerry.Mr. Chairman, commissioners, Mr.
Gore mine and members of the DAC committee thank you very much
for inviting me to speak here today. My name is Jerry Brito and I’m
the executive director of Coin Center, the nonprofits in DC that’s as — on
cript I had an opportunity to present
to the committee about the first can —
about bitcoin and currencies, the the
neagh focused on what was bitcoin and blockchain,
technology, how it worked and what were social and regulatory
implications. We’ve come loong way since then
we’ve seen cryptotechnology and
networks grow and catch the — in a big way. We’ll give you an update on
dotle we’ve seen in the technology and public policy in
the three years since that first meeting. So let’s start with
the technological developments. The most obvious change looking
at the landscape of cryptocurrency today is looking
at the cryptocurrency network and tokens over the past three
years. Two years ago there were a small handful of blockchain in
existence with bitcoin being the gorilla with
almost 100 bers market share since then many cryptocurrencies
and tokens have been developed and today there are over 1500
tracked by market tap which is an industry website.Some of
these are essentially copies of bitcoin with few in if any
improvements or difference toss distinguish them. Others are
remarkable efforts to extend the break throughs developed in
bitcoin, to accomplish new things and allow for new
applications. The Etherium launched two and a
half years ago are one of the most
promising currency projects to occur since bitcoin and today it accounts for 20
percent of the market cap of cryptocurrencies. Whereas it’s center simple and conservative to be the for most
peer to peer electronic cash the
theories of fine gold is ambitious to be a decentralized global computer on
top of which anyone can run a number of applications from
personal identity systems to en assurance crowd sourced hedge
funds in dead of the 1500 token projects I mentioned earlier
about 600 do not run under own blockchain network but exist on
top of Etherium. Another sed of new
cryptocurrency networks are those like Z cash
monero and dash which aim to improve bitcoin’s design by
eating privacy protecting features. Privacy and
fungibility are two of the most salient features of cash but are
missing from bitcoin. These new cryptocurrency
networks use advanced cryptography to allow
curms to keep transactions private and selectively disclose
these transactions as they wish or are compelled to
through legal process or required through regulation.It
also turns out that private transactions are very important
to financial institutions who have been experimenting with
their own DLT solutions, and have found that simple blockchains like Bitcoins are
too transparent. Projects like JP Morgan’s quorum
an enterprise focused blockchain solution that is looking to
incorporate technology from Z cash to keep transactions
confidential to the involved parties, but verifiable
to the larger network.Finally, how these networks can scale to
accommodate the thousands of transactions per second that
would be necessary for global adoption is a major technical
challenge that has been one of the main focuses of developers
over the Bitcoin today tops out at about
10 transactions per second, but employing second layer solutions
like the lightning network or the rating network on Etherium can potentially
allow for massive scaling by allowing for secure offchain
transactions and only using the blockchain for settlements.This
technology has been developed over the past couple of years, and
the first working networks launched this year. Let me turn now to developments
in public policy over the past three years. The top issues
facing cryptocurrencies are consumer protection, securities regulations, tax policy, and
anti-money laundering regulation. Let’s take those in
turn. Regulation of actors in the
cryptocurrency ecosystem to en sure protection has been
primarily through state by state and money licensing the policy
rational here is if a business is taking cufrsd of consumer
funds even if it’s just momentarily in order to provide
a financial service, they pose a risk of loss to consumers. To address that, they must first
receive percentages from the state, typically after passing a
background check, posting a bond and meeting other requirements,
before they can engage in business. Two problems emerge
with this state by state regime. First is that firms must seek a license in every state in which
they do business in, which they have customers, which for an
internet business means every state. Even though ttion made
no safer, the firm is made no safer to consumers when it
passes its 50th background check than Not only is this inefficient and burdensome regulatory impediment
to inherently interstate commerce the state regulators
have been slow in interpreting how its licensing requirements
apply to cryptocurrencies.Second because these laws were
typically written decades ago they are not written in terms of custody which is the
risk basic to address, but instead in terms of
transmission. Because before tin vengs of
bitcoin, custody was inherent in transmission. Cryptocurrency networks however
allow one to perform many different activities that could
be construed as aiding the transmission, and thus
potentially covered by licensing
requirements, but which never involve custody, and thus a Operating as a minor or a node
on the lightning network or offering a noncustodial wallet
service are all examples.Just under three years ago the state
of New York created a technology-specific license
called a bit license, which was — which has been universally
panned by the cryptocurrency and legal communities. Since then
other states have amended or interpreted their statutes with
varying degrees of success. More promisingly, the uniform
law commission has developed a uniform virtual currency
business regulation act that is excellent and is now being Unfortunately a state by state
reform don’t scale.State by state money transmission
licensing additionally has no provision for market supervision
or exchanges, which is increasingly of interest to
federal policymakers. I was therefore happy to see a
CC chairman claip ton and Giancarlo
say they would support policies to
— the state by state regime. How these laws apply to
cryptocurrency is the next pressing policy question that’s
emerged oist past years. Many of the tokens that have
been proliferated grynl have resulted in ICOs to investors
who buy them with the expectation of profit based on
the efforts of the party that is selling the tokens, and their
promises to build infrastructure through which the tokens will
eventually have utility. Such tokens are clearly
securities, and the FCC through investigative
reports and enforcement actions have begun to make this quite
clear. It’s important to note however
while testimony tokens are securities not all are and I was
gratified to hear chairman Clayton several times in a
recent hearing draw the distinction between
cryptocurrencies on one hand and tokens that have been issued as
securities on the other. Cryptocurrencies like bit cones
are commodities of course as the SEC has previously sound. quegz
questions remain however about the border then these categories
and about how one can responsibly share tokens to
future investors. Anti-money laundering was the first area of
regulation intersecting with cryptocurrencies that prompted a
federal response in the form of fin sens
2014 virtually guidance it’s one of
the most stable and policy areas, the rules here are pretty
clear, if you are a cryptocurrency exchange or
similar intermediary, you are subject to
the secrecy act and obligations all basic exchanges comply with
these rules over the past three years we’ve seen
fin sen issue supplementary guidance and
freeway greater authority offshore rogue changes continue
to be a concern to law enforcement. Finally on tax the IRS issued in
March of 2014 guidance finding that cryptooccurrences like
Bitcoin are treated as property for tax tax purposes which meanings gales for sell or
gain are capital gains rather than ordinary income however
unlike traditional currencies property does November a de
minimis exemption law. This means that each time you buy a cup of coffee with Bitcoin it is
a taxable event and you are technically required to record
report and pay taxes on any gain in a you experienced even if
it’s just a few pennies. This obviously introduces
friction that underpipelines potential for using
cryptocurrency for day-to-day payments or micro transactions. Last year representatives For personal transactions under
$600. It’s a lot like what exists for foreign currency. There are many other policy
questions that remain open in tax and other areas of law, but
those are the highlights. And I’ll stop there, and I look
forward to your questions. Thank you.>>Thank you. Mr.
DeWaal.>>Thank you, Dan. And thank you, Jerry, for completely
anticipating everything I was going to say, but I am going to
completely speak So last year at the FIA law and
compliance conference I had the privilege of leading the kickoff
panel with Kimberly John, and it was entitled memory lane, those
who forget history And I’m going to take one brief
diversion to be talk about something that probably occurred
before most people around this table became part of the
industry. And I want to take you back to
the 1970s. In the early 1970s, the CFTC
didn’t exist, the comoth commodities
exchange authority oversaw this business, and options on
enumerated domestic grurchl commodities were bad. There was
need for these products, hedgers, commercial users, had
reason to want to use these products, but the products were
general banned in the United States. Commodity markets became very
very volatile in the 1970s and there became quite a market in
world commodity options and they were traded in
London, and unfortunately some bad guys decided to start
offering and sell them in the United States. Caused quite a
problem. Bad firms, Goldstein Samuel son
ended up being put out of business,
folks lost their money.And frankly among other reasons,
this is why this agency began. The commodity exchange act was expanded the definition of the
commodity exchange act was amended the
definition of commodity was expand to society included not
only the additional enumerated comod commodities
under the overside of the Department of Agriculture, but
moj other things plus it gained the potential for expansion. And the CFTC took that authority
when it began operations and it used that authority to go after
the bad guys who were dealing with the commodity option space. Sadly, despite bringing
something like 55 injunctive actions between its formation in 1978, it couldn’t
keep up with the bad guys. Moreover, there was confusion
out there. What were these products. Are they securities, are they
commodities, are they — sound familiar? Well, the reaction then,
fortunately, is not what this agency is doing today. For three years this product was banned in the United States by
the CFTC, And it was not until a pilot
program was reannounced in 1981, and then some of the exchanges came around and
proposed specific contracts through trade on designated
contract markets, that options trading began again in 1982 .
Now, all these years later, nobody thinks about commodity option a
problem, these are valuable tools used daily by commercial
users, by producers, by hedgers, they’re traded on
exchange contracts, regulated by the CFTC. They’re traded also
OTC. And no one thinks about them as
a problem, they only think about them as a solution. But once upon a time, they were
seen as a real problem. We now fast forward. Bitcoin started trading in 2009,
I’m not going to go over again the evolution of laws, I think
Jerry did a great job discussing the situation out there right
now. But my concern is that unless
there’s some kind of rationalization
process, it will exceed the development of
not just coins. Coins are critical to the
decentralized ledgers. They don’t exist in a vacuum. They are the mechanism in proof
of work blockchains where minors
are rewarded, in proof of state
blockchains where fees are paid these are the way you
incentivize folks to keep the system together. If you’re only
talking about centralized ledger sure youf you don’t need to
worry about coins. But if you think there’s validity in
decentralized ledgers, and you think these guys that are work
on the daps, that are sitting on thop of the ethereal blockchain,
yeah, may have some good ideas. These have to be thought about
carefully.I just want to say a couple things about the
legislation, about the state of legislation. The states absolutely began with
the money transmitter rules, and that’s predominantly where most
of the states are today. New York state came up with a bit
license, which I agree hasn’t been warmly received. Six I
think there are six folks out there right now that either are
regulated as limited purpose trusts effectively applying the
bit license Last week New York state said
you know what, we’re going to expand the requirements of
people with bit license, and we are going to get into the
requirements dealing with, you know, oversight and manipulation
and fraud, et cetera.But it’s not like the bit license
replaced the money transmitter requirements. It’s both. And
it’s not like anyone is talking about implementation of this new
virtual currency act replacing money transmitters. Maybe it
will happen, but we’re talking about potential additional
regular laigz lation at the state level.Oh, and by the way,
if you haven’t caught the news in the last couple of weeks, Massachusetts, New
Jersey, Texas, and North Carolina, the state
security commissions have started bringing action gains
so-called prohibited I cros where there wasn’t registration
of security so we’re now seeing a third branch potentially of
states getting involved in this space. I just want to spend a few
seconds on the SEC’s oversight there’s a very famous case you
hear it all the time it’s mentioned in the SEC enforcement
action the howy case, and it talks about the fact that something is an
veaflt contract namely a type of
security, if it is an investment, an enterprise with the expectation of profits,
through the managerial or entrepreneurial efforts of
others. Howie actually uses substantial
efforts of others but SEC over time has just talked about the
managerial or entrepreneurial efforts of others others.That’s
been a rule that the SEC has relied on to recently go after
the mun chea, dap, and is using to go
exercise jurisdiction in some other ICOs
and prohibited securities. And I agree in most circumstances, a
lot of the stuff out there absolutely looks like a
security, something needs to be registered or issued when it’s
But not 100 percent. And if you think about the application of the howie test to common
things suppose you bought a Tesla, one of the first Teslas
out there and you bought it because you know that musk was
going to be out there pushing this thing and saying it’s great
it’s wonderful it’s exciting. Well sure I’m buying it, but
there’s not too many places I can actually plug in my Tesla. I’m getting it for the
investment value. I know that’s first edition.Is that car a
security? Is that car a security because I’m expecting
somebody else to promote it and there’s hype going on about it? Privately issued gold coins.
People buy them, people hold them, people expect to resell
them for a profit. Are those securities? Because the minters are
promoting those? Ultimately there has to be some
clarification. The distinction between a commodity, the
distinction between a security, may seem from a common sense
perspective clear, but there are very, very important issues
around those that I think this committee could very much help
clarify. And move the regulatory
discussion along. And hopefully that’s something
that we can do, and I certainly would look forward to
participating in that.>>Daniel Gorfine: Thank you. Mr.
Gorelick.>>Richard Gorelick: Thank you very much, and let’s
see if we can get this presentation working as well. Anyone here who works on — — okay, well, while we’re
getting this going here, I’m going to do just a brief
swrouction, I’m Richard Gorelick I’m the head of market structure
at DRW Holdings, I appreciate the opportunity to participate
in this important dialogue as a member of the technology
advisory committee. I’ve been on this committee
since it was reconstituted in 2010 and I welcome the commission’s initiative to
assemble it I was previously a co founder
and the CEO of RCM advisors that is an
active participant in electronic markets. Since DRW acquired RGM trading businesses in September, and I
joined DRW I’ve had the opportunity to learn about a
variety of new markets including this rapidly developing market
for cryptocurrency such as Bitcoin. The commission has asked me to
share a little bit about the market structure for the cash
and futures markets for cryptocurrencies today, and
that’s what I’m hoping to do momentarily here. Let’s see. And as you pull that up at least
for members it is in your materials you’ll see a printout
of the presentation which we’ll hopefully have up on screen in a
moment. There we go thank you very much. So I’ll start off by noting I’m
using the term cryptoassets and this is intend to be a real
broad term. A lot of the time we’re talking about
cryptocurrencies but as it turns out, a lot of these are not in
fact something that we woud widely regard as currencies,
have very different characteristics, some of them
might be in fact cryptosecurities as we’ve talked
about, some of those might be utilities tokens or other things
I’m trying to use DRW has been active in this
cryptoecosystem for a number of years. DRW recognized Bitcoin and the
underlying technology as an opportunity in 2012. Cumberland was founded in 2014
as a subsidiary in a bitcoin trading
desk which has been uniquely positioned between traditional
financial issue where DRW has participated for a couple of decades and the nascent
cryptoasset place. Today couple ber lan specializes
in providing two sided institutional sized liquidity to
parties around the world. Cryptoalso manages a long
portfolio or cryptoassets and pursues
strategic opportunities in these emerging markets. So that’s why we’re involved and
why I’m here today to talk to you about it.I already spoke a
little bit about terminology. Bitcoin was the first cryptoas
set lawrchld in 2009 and today as we’ve heard over 1500
cryptoassets exist. The volumes, the trading volumes,
and the market caps of these coins
are still skewed toward Bitcoin and a small handful of others as these
charts indicate. Bitcoin, ethereal, bitcoin and
cash are the top both in terms of market cap and the volume
drops off considerably and market cap beyond those. In terms of the size of the
market, the market has grown considerably over the last year.
This chart starts in January of 2017 and what you can see is
that the overall market cap has gone from approximately zero at
this scale up to as high as over 750 billion in December, and
recently settled in the 400 billion plus range over the last
couple of days. At the same time, the trading
volumes of these coins has gone up from approximately zero on
this scale to between 30 and $60 billion of value per day. And so because of sort of the
shape of these curves and trend in both increase in market cap
and volume particularly at the end of last year that’s why
there’s been so much attention on this space in the last
several months.Another thing to note in terms of market dynamics
over the last year and a half as Jerry discussed a little bit, is
that at the beginning of last year, this was a market that was
completely dominated by Bitcoin, over 85 percent of the market
cap for cryptoassets was in Bitcoin at the beginning of last
year. Over the course of the year there was much more
diversification. As much as we talk about the
major runup in Bitcoin prices last year, it actually
underperformed a number of these other coins, so the result
is by in curndd percent about 35 percent of Want to talk a little bit about
the trading conventions for cryptoassets. It’s truly a 24/7 market. It
trades all night, all weekend, it trades on holidays, it just
trades around the clock. Other markets like FX talk about being a 24/7 market, but do take some
time off occasionally. This is really around the clock. A
couple of the other interesting dynamics very small pricing
increments. So generally for a Bitcoin which trades in the let’s say $9,000
range as of today, the tick increment, the price increment,
is one penny. So it’s much smaller than a lot of the
markets on equities and futures side that we’re used to for
assets that trade at that price level. And at the same time, the lot
size is very small there’s really essentially no limit on the smallestize that
you can trade of a Bitcoin. Right now it is a spplt — point
7 zeros and one of a Bitcoin also
known in satochi it’s 11 00 millionth of
a Bitcoin can side often you see
very small sizes on the trading screen the combination of these
tiny pricing increments and these very small lot sizes for
trading in the spot market means that you have a highly fluid and often very thin order book on
exchanges that trade.I’m going to talk a little bit about the
spot exchanges. There are over 100 exchanges
around the world that trade bit appoint and other top
cryptoassets in what I’m calling the spot market. And I’ve put
exchanges in quotes because it doesn’t really fit into the
regulatory framework of how we traditionally think about
exchanges in more mature and more regulated marks like the
futures markets and the equities markets around the world. The largest exchanges include a
number of the exchanges around the world, and I think the
important thing to note here is they’re not all based in the
United States. Bit finex is based in Hong Kong,
bit stamp was founded in Slovenia and subsequently moved
to UK and Luxembourg, as you go down the list you’ll notice there’s Japan, there’s
Singapore, various UK and European presences, thtion truly
sort have of a global market. And as I mentioned our trading
desk Cumberland, we’re an active participant on a number of exchanges that
have meaningful trading volumes. The Korean changes received a
lot of attention in the last couple months and they’ve grown
in importance. In December it was reported that about 20
percent of the world’s yip to asset volumes were in south
Korea, and this was the third biggest market in the world
after the U.S. and Japan. New regulars went into effect at
the end of January that had the impact
of slowing that down a little bit, but we’ll have to see,000
how things play out over time but a lot of discussion about
the Korean exchanges in recent months. Generally speaking I think
trading on these can be challenging particularly if your
goal is to trade across multiple spot exchanges it’s difficult to
weave toygd liquidity across exchanges and jurisdictions due
to a number of factors. Technology across these
platforms is very noncentered particularly if you’re coming
from the traditional industry these don’t look like normal
protocols we’re used to. There’s concerns about deceptive trading, like wash sales and
spoofing, and other types of manipulation. There’s not
standard best practices, every exchange does their own thing in
a different way and we’re really behind the ball in sort of
developing standard ways of doing business with these
exchanges. Banking relationships have been
fickle, and what’s happened here is that exchanges have had hard
times getting banking relationships in various
jurisdictions, sometimes they get them and then they lose them
and they move them around, and that has created difficulty at
times moving money in and out of these venues. Because of the need to post both
money and coins at all of the exchanges on which you want to
trade it’s very capital efficient to trade across
multiple exchanges.There are obvious well known concerns
about cybersecurity and concerns about the transparency at the
exchanges about who runs these exchanges, what their financial
wherewithal with, what their practices are, and what exactly
they’re doing on exchange. And it can be very slow to move
money and coins in and out of these
various exchanges. On top of that, the fees are higher than
we’re used to in sort of more mature financial markets. So
there’s a lot of challenges in dealing with these spot
exchanges. Another thing to note is that it
often appears that the markets are crossed. Meaning that it
looks possible that you could just buy on one exchange for
less than you could just sell on another exchange. The Korean exchanges, for
example, had a period of time that sort of mostly gone away, since the new regulations
have gone into effect at the end of January, where the pricing on
the Koreaing changes for Bitcoin was much higher than the prices
on other markets. At the bottom of the screen you’ve seen an
example of just a quote that I pulled a few weeks ago, where on
bit Finn ex Bitcoin was trading point $6,000 and major Korean upexchange, the
it’s treading $15,000 in theory crosses like this should be eliminated by arbitrage and
people say wow there’s a lot of money to be made by arbitrage
but there are structural reasons for these price difference
including exchanges and delays and difficulty on getting some
of you these exchanges and that needs to be kept in mind. Algo trading is another topic
I’ve been asked about. It is developing in pockets of
the cryptomarkets. But most exchanges today are hosted in the cloud, and as a result low
latency is not yet a significant differentiator in these markets. The latencys the times we hear
about to execute are seconds to tens of seconds range at this
point while people try to speed it up in areas, that’s not a
dominant feature of the market yet. Over the counter trading in this
market is very important. A meaningful portion, I’ve had
difficulty on getting firm estimates of exactly how much,
but a meaningful portion of all crypto asset
market volumes are occurring away from these exchanges. Many are with over the counter changes, DRW Cumberland trades
over the counter markets in institutional size. I think the
advantages is for some market participants of trading in these
over the counter markets is that the ability to trade in large
institutional sizes, the abilities to get competitive
all-in pricing, and the ability to quickly settle and get paid,
usually in less than 24 hours, are all features of the over the
counter market that may be advantageous relative to the
sort of difficulty to weave together exchange space. Bitcoin futures launches. The CFD launched fwoyck on
December 10th CME followed on December 17th our exchanges are likely to launch
cryptoasset based in coming months the creation of these
products is a real positive for cryptoasset industry. It
demonstrates the overall maturity of the business as an
asset class it’s very good to have established and regulated exchanges involved
with the landscape. Our view is that ation new products are
launched in the future and as new exchanges get involved, we
would like to see a good healthy dialogue
between the exchanges in bolt the financial community
and cryptoasset community that will trade and participate in
the development of these markets.Challenges remain with
the supplement mechanisms, the first rounds of settlements
wents relatively smoothsly but we continue to have concerns
that the way that these futures contracts are pegged to these cash markets,
which are less transparent, could result
in dislocations in the future, and this is something we’re
watching closely and hoping to work with the exchanges as
things go bard forward. We’ve expressed our view that we would
like to see physically settled cryptoasset contracts to help
deal with some of these concerns. We recognize that
that may be a little quick comparison to have CFTC
and CME in terms of market structure there, the CFTC is one Bitcoin whereas 5 in
the CME this gives you an idea of the size. The open market is — CME is a
little bit larger if you look at it in terms of Bitcoin terms.
Daily volumes are comparable at between 5 and 15,000. Bitcoins
worth of contracts as each exchange. Bolt are cash settled with
different settlement mechanisms, the CFTC mechanism relies on the auction
price, while the CME settlement relies
on price averaged across four spot exchanges at present. Tick increments are much bigger
than we see on the cash market on CFE
it’s $10, and $5 on CME this exairtion to
one In terms of foreign regulation,
chairman Giancarlo has gone through in some of his recent
speeches a lot of the regulatory approach is overseas is very
important I think the important thing to note is these are truly global
markets, it’s truly a global asset class so the ability of
any one regulator in any one jurisdiction to do a lot in this
space is limited by that sort of global nature. That’s not to
say that we shouldn’t try, we certainly should. But we need to to do so in a way
that’s respectful there are other markets and other
jurisdictions where Governments around the world are trying to figure out how to regulate
Bitcoin and other cryptoasset with various approaches, China
banned some exchanges in September. Ace mentioned, recently the new
regular laitionz in south Korea are trading through what they
call real name bank accounts to improve on some of the anti-money laundering and KYC
concerns. Japan has been a thought leader
in these spaces in a require exchanges to register, for
authorization and last time I checked there were over 15
exchanges that had registered and were being supervised under
this new France recently announced a
formation of a task force to propose a new framework My other members of the panel
here have dob A done a good job talking about U.S. regulatory
guidance while I concur that there’s been a lot of guidance
and investor education, in recent weeks and months from the CFTC and the
SEC, there’s still room for improved clarity around what the rules are, particularly
around the definitional what is a
security, what is a commodity what is an asset — what are the
rules and how do we determine what is what. My overall proposal for the
regulatory approach is that the UC should adopt a smart
principles based regulatory framework that encourages
professional and responsible market peapghs to build To do so, the market is going to
require more certainty as to the legal status of these cryptoasset, better
guidance about how financial institutions can provide services to the cryptoasset
businesses, and zero tolerance for fraud, scams and abuse. And I commend the CFTC and the
SEC for recently making — getting very involved in this
space, and talking about their view that that’s their primary
focus, is to make sure that these markets are safe I’m going to suggest the TAC
working group to look into these issues in depth to make specific
recommendations to the CFTC consistent with this approach.
Thank you.>>Daniel Gorfine: Thank you
very much, Mr. Zaidi.>>Amir Zaidi: Thank you. Thanks for
having me today my name is Amir Zaidi I’m the director of
the division of market oversight here at the CFTC and I’ve been
asked to talk a little bit about a related
aspect to this whole conversation, is our
recent retail commodity transaction
interp specifically the actual delivery exception to that — to
our jurisdiction over these retail commodity transactions. And more specifically, with
respect to virtual currencies. Before I start, I will also give
the standard disclaimer that these are my own views, since I see our
general counsel staring at me. As you all know, there are
several well-established platforms in the U.S., as my
panelists have gone over, engaging in U.S. retail investors in a growing
number of virtual currencies that are available for trading.
As previously noted by the commission, the underlying cash marketplace
for the virtual currencies remain largely unregulated, and the CFTC has
limited We do have anti-fraud, general
anti-fraud authority, the commission also has statutory authority to
treat these certain leveraged margin
or finance retail commodity transactions as if they were a
commodity futures transactions. This means that certain virtual currencies platforms offering
these retail commodity transactions to U.S. retail
investors must register with the commission as DCMs. This also means that trading in
such transactions may also trigger commission registrations
for certain intermediaries Like I said, the commission’s
jurisdiction over these retail commodity transactions includes an
exception when actually delivered to the purchaser, actual delivery to the purchaser
occurs within 28 days from the date of the transaction. As per relevant commission and
jurisdictional precedent, a finding of actual delivery is
focused on possession and control. Accordingly, the
proposed interpretation attempts to properly recognize when a
purchaser has sufficient possession and control over
virtual currency in retail commodity The main point of this proposed
interpretation is to clarify that actual delivery of a virtual currency
to qualify for that exception from
the commodity exchange act requires purchase freedom to
utilize the commodity purchased no later than 28 days from the
date of the transaction. At this point, the seller and
any platform involved should no longer be able to prevent use or otherwise
take away any amount of that virtual currency purchased using
leverage or margin. The commission notes that it
drafted this proposed interpretation with a balance in mind, to avoid
impeding upon market enhancing innovation while ensuring
integrity for U.S. retail investors, when the transaction
falls within this commission’s So the baseline interpretation
proposed interpretation contains two primary factors required to
demonstrate actual delivery. The customer having the ability
to take possession and control of
that entire quantity of the commodity, whether it’s
purchased on margin or using leverage, or any other financing
arrangement; and the second, to use it freely in commerce, both
within and away from any particular platform no later
than 28 days from the date of the transaction. And secondly, the offeror and
the counterparty seller, including
any of their respective affiliates or other persons
acting in concert with the offeror or counterparty seller
on a similar basis not retaining any interest in or control over
any of the commodity purchased on leverage, margin or
other financing arrangement at the expiration of 28 days from
the date of that The interpretation goes on to
note, four examples to further elaborate on
the baseline in this interpretation in example 1 it requires a full
blockchain audit trail demonstrating that
the to the vrch currency of the seller within 28 days of the
transaction, when a platform is involved to intermediate yeat
the transfer would transfer to the virtual currency to the
prrcher within 28 days. In example 2 there’s a little
more flexibility allowing certain virtual platforms to
offer wallets for use by their customers and act as the
depository if the purchaser agrees while ensuring that the
platform ultimately gives up any control over the purchase of
virtual currency once that 28 days has passed from the date of
the transaction. In example three provides that
mere book entries are not enough to demonstrate that actual
delivery occurs, something more will be necessary to confirm to
the — conform to the baseline interpretation. And then the final example
explains that actual delivery must have actually occurred at
some point. This means that actual delivery
is not satisfied if the transaction is simply rolled,
offset, netted out or otherwise We also have — obviously this
is a proposal so we’re asking for comment. The commission
asked multiple questions to the public to further refine the
interpretation. A few examples, the commission asked about the meaning of certain
terms such as title, depository control. Further, the
commission asked about the adequacy of that 28 day delivery
period. The restriction on liens beyond
the 28 day period and the potential for a unique
regulatory regime for these types The comment period is still
open, ’til March 20th, I believe, so
another month to get your comments in, and we’d appreciate
those and look forward to reading them. Thank you.>>Daniel Gorfine: Okay great
thank you like to thank our panelists you walked through
market development, existing legal regulatory questions and
challenges, trading activity, and then a view from the CFTC.
So thank you. I want to open it up again to
questions and discussions. You can start thinking of comments,
and again please place your card upright if you have a question
or comment.Before I do that, though, I’d like to actually
turn to Mr. Durkin and Mr. Chu to talk a little bit about
recent updates with respect to Bitcoin
futures and Bitcoin options trading that’s been taking place
on your respective exchanges. I’ll start with you, Mr. Durkin.
>>Thank you Mr. Gorfine, and I just want to
commend the commission for its efforts and leadership in the
successful launch and introduction of our Bitcoin
futures contract, also want to
complement Commissioner Behnam for his discussion at the MRAC
on this topic a few weeks ago. To bring you up to date, as the
world’s largest futures and regulated marketplace we did
feel that CME was a natural home for the establishment of our
Bitcoin futures contract.One of the key strengths of our
offering was predicated on our already
successful Bitcoin reference rate, which we introduced in
November of 2016. And that reference rate was
introduced to provide greater transparency
to this spot Bitcoin market. Now, the Bitcoin reference rate
is a once per day reference rate based on the U.S. dollar price
of Bitcoin. And it aggregates as was alluded
to here, data from four exchanges,
be it bitstamp, GDAX, E bit and
crackin, that demonstrates reference rate for buying and
selling Bitcoin and its brought into an aggregate of an order book
which reflects real time again U.S. dollar price of Bitcoin. We were pleased that we were
able to introduce that some 14 months
ago, which then brought us onto the launch
of our contract. So the contract was launched on
December 18th, and to date we’re averaging approximately 1,397
contracts daily, which translates into 83
million notional. 65 percent of that volume has
traded during U.S. hours, 35 percent of that volume is traded
during non-U.S. hours. We have over 900 unique accounts
that are routinely trading across all
of our clients segments. Mark maker volume has been
approximately 27 percent of the product since launch. 73 percent of the volume is
nonmarket-maker customer volume. Retail customers have
represented approximately 13 percent of the volume. The buy
side is representing approximately 15 percent of the
volume. Props around 70 percent, and banks a little bit
over 1 percent. Our open interest is 1,936
contracts, or the equivalent of 81 million in notional. And
that open interest is growing each day. And that has been the
case since launch. We have very active quoting
volume and interest across all of the
expireees, which is very exciting in the context of the
buildup of the contract. And we’re very pleased with the
risk management associated with this contract. So it was very fundamental to us
that we implement a variety of risk management tools such as
higher margin levels, position and price
limits, credit controls to appropriately manage and
aggressively manage the risk of this contract. And I’m very
proud and pleased to say that every single aspect of that has
performed according to standard.>>Daniel Gorfine: Thank you.
Mr. Chu, and I believe you may have some slides that perhaps
Amir is going to >>Thank you Amir really
appreciate the assistance on this one. First I want to say
that the agenda is really fantastic, in that obviously the topics are very timely, but also
very synergistic. If you deal with things like
distributor threrj technology in Bitcoin obviously cybersecurity
is going to be top of mind as well. So I hope as we go
throughout the rest of the day we can bring back some of the
topics that have been discussed this morning in a way that those
thinds things kind of interface my name
is Paul Chou, my wife, happy valentines
day. We started LedgerX four years
ago and we’re currently a swap execution facilities and
derivatives clearing torsion registered with the CFTC. And our focus is is entirely on
cryptocurrencies and in particular what we folks on is
derivatives around physically settled cryptocurrencies so in
this sense you actually get the underlying Bitcoin or other
digital currency that the contract actually references. So this first slide up here, you
know on the surface ledger X looks no different than any
other clearinghouse operation you might imagine. We take
dollars for sure, and our specialty is we also get crypto,
that’s actually pledged to the clearinghouse to underlie a lot
of derivatives trades that interacted on our SEF.When we
talk about things like physical delivering and settlement what
we mean is they actually send it to a clearinghouse address that
only we control. So the DCO at that point has
full positive control over all the crypto that we manage, and
an example transaction might be for example a customer comes in, wants to buy
a call option, they send U.S. dollars to us. We have a
customer who wants to sell a call option against Bitcoin,
they transport a Bitcoin to DC entity we hold it for it and
expiration of the call option if the customer chooses to exercise
then we take the strike price give the call seller and
actually deliver in Bitcoin. That’s a really key point for us
because a lot of our customers need the underlying Bitcoin for
a lot of operations. So you know to note I think
Bitcoin and other digital currencies have like a really
great future in terms of collateral, it always struck us
as a great collateral instrument. You can send it to us 24/7, 365,
you don’t have to worry about banking hours. In fact a lot of
our customers send Bitcoin on Sunday at 4 p.m.. And within a
few hours they’re actually able to trade with that collateral in
hand. So I think it’s a very powerful use There are certain
customers of ours and clearing members that don’t even deposit
dollars at all. They can send Bitcoin to us, through all
manner variety of swaps and transactions and options, and
then when they’re done convert the Bitcoin and actually
withdraw it to their personal So this I think is a very use
case that cryptocurrencies can do that others can’t,
essentially not touch the U.S. banking system at that point. It’s something we think we do
very well especially at ledger X
because the physical accessibility is quite
important. Next slide, please. I want to go over a little bit
the different players in the eexo system. There are a lot of
things that are really not that different from normal commodity
markets. Bitcoin has naturally loans, has
natural — longs, natural shorts has long term investment managers
and short term traders as well I was trying to emphasize to people we can’t use
a lot of the technologies with a lot of markets for the
participants that do work in the Bitcoin system, there are many. A classic example, like a long
holder, natural long, would be a minor
so this individual or institution has spent many years
investing infrastructure, designing specialized chips
paying for the electricity behind that and also employing
people, and they get commodity as the result. So it’s not too different from
British petroleum, which get oil
commodity out and have they have those price
risks as well. What a long holder might do is take Bitcoin
they’ve mined over the years, and do call rights against them
that’s the classic trade for this sort of thing we get U.S.
dollars today we get Bitcoin as collateral against those call
rights and use U.S. dollars to use research and development for
generation of their chips or expand their factories things
like that. Long holders obviously are very interested
because Bitcoin is potentially like an uncorrelated
asset class, like other asset classes
what we’ve seen in financial markets,
they tend to be moving a lot together more than people would
like and adding some portion of your portfolio to crypto, could be a very valuable
addition to Finally money market make ertion are obviously
attracted to things that move, and things that have quite a bit
of volatility. And Bitcoin and additional currencies we have
volatility in space. It moves a lot. And so I think there are a
lot of traditional people that used to trade in Nat gas or or other comods
commodities that are looking at crypto, and
way toss capture or spread around that.This is a particular
trade I’ll highlight one of the things we’re seeing in the
market that was covered by various news outlets and
basically it was done from our public reporting. There was an
individual that bought 1 million dollars worth of premium, so
that’s not a $1 million position, it’s a million dollars
worth of call premium. That Bitcoin by tend of December
of this year would be about $50,000. So you know we’re starting to
see some distributional views that very sfraiscated market
participants have. So to give you a sense of scale, if Bitcoin
were to end that $50,000 by the end of this yearks the entire
market cap with Bitcoin, not any other
digital currencies would be about a trillion dollars. A advisor would look at this,
we’re starting to have oversight over something that could be one
day a trillion dollar market cap entity I think it’s important to
put that out More importantly, options that’s what we specialize in also give us a
view of what market participants think is the riskness of Bitcoin
going for. Even if you have no intention of actually trading
options on Bitcoin, say you only trade spot Bitcoin, this gives like an unprecedented tool of
market forward expectations as to how much it’s going to move,
and therefore risk managers are looking at the swap positions,
can scale their positions That’s something that visibility I
think has been kind of lacking for quite some Next slide,
please. All right, so I talked about some of the things that
ledger X is seeing in terms of trading. What I want to mention
is something that all these unique things about Bitcoin that
we’re seeing in clearing. So most of the people here in the
committee probably have head of things like hard forks, Bitcoin
is software, so every once in awhile somebody develops a new
version of software, and if you’re holding Bitcoin in
November, you might be holding two different — two coins in December, based on the hard 4.
So this graphic up here is basically all the hard 4s that
happened for Bitcoin in December alone, so if you can see if you
held one CPTC you’d have a platform of other Bitcoins down
the line it’s almost like a dividend in some sense but it
provides a lot of unique interesting challenges for a
clearinghouse. Because if you list a put option
for example on Bitcoin and it expires in a year and put option
holder wants to exercise it, and put the underlying to it, what
are you asking them to put. Was it just the original coin or
is it some subset of the coins that forked over the last couple
of years. So we have as you might imagine very rigorous
conversations about this exact issue with our clearing members.
Some choose to just withdraw their Bitcoin from our
clearinghouse before fork happens, some don’ts believe
that the forks will actually have value, so people are
constantly doing different strategies to manage this
particular risk but I want to highlight this as unique to
cryptocur currencies and also something that’s unique for
industries that develop best practices that
technology advisory committee is uniquely suited to establish and
establish best practices and industry status.>>Thank you Mr.
Chou. I do want to open it up now to questions or comments or
statements from our members. Again, with an eye towards what
are some potential work streams that this committee may
consider. Let’s certification I see Mr.
Curly.>>Thank you very much Dan for having me on the panel and
to the commission as well. So I wanted to focus on a couple
of points that are refensed at least in passing by the panel.
The first is custody. And custody is something that is
really taken for granted in many other cases but really here
fundamental to both consumer protection and to the
functioning markets. So that’s something that we definitely see as an area that could merit
some attention by something like the TAC in the following sense,
that there are differences between what custody means for
different types of asset. So we experience a lot of people
ask questions about why am I
becoming a bank for this reason, it’s because
there is a recognized regime for trust
companies or banks that can perform custodial functions, but
the observation oftentimes for these parties is
really they’re proposing to do something very different than
what the banking supervisor would otherwise be asking them
to do. They’re facing technology
challenges different because of the nature of their assets or Holdings and it’s
often a very gers diverse set of assets
there’s dircht 100 different boyction and assets they want to deal with therefore
it’s a problem with custody, a security
cash or other types of assets that have more traditional types
of structure. That’s something I would invite
the panel’s view on whether they’re experiencing other
things, and secretary as — second aspect being related to
ildd and that’s accounting, accounting being a challenge
that we see as parties facing pretty regularly in terms of
what the accounting standards are, how to deal with the keys without
undermining the anonymity or other features of the platform. But again fundamental to
assuring that assets are there when they’re expected to be.
And that the value that are represented in derivative or
other markets is tied to something that’s there in fact.>>To custody what I would say
is it is a question that so many different laws and requirements
are predicated on And that’s an open question of
what constitutes custody, I think part of the What I would recommend to the
TAC is that they look at the uniform
law commission’s virtual currency regulation act which
has little to do with what you’re doing here however this
committee of the ULC spent two years thinking about this. The licensing in that act is —
turned on custody, and they developed a definition of
control. Is what they call it. And we think, you know, it’s the
best that we’ve seen. So I would recommend that you look at
a definition of control in the ULC act when thinking about
custody.>>Daniel Gorfine: Okay, I’ll
go to Mr.>>Thank you, and thank you to the panel for a great
presentation. I would recommend that the TAC
consider a working group to look at regulatory rationalization in
the space, and one area, one sort of deeper question I think
we should wrestle with, is given this sort of flexibility and versatility of virtual
currencies or assets have shown, and the fact we’re wrestling over the terms
indicates it, it’s like the story of the elephant and the
three blind men. You’re looking at the — the regulators are
looking at it from different perspectives and thinking it’s a
different thing and you find this elephant regulated by the tusk,
tail, and trunk regulators all at the same time. And so one thing we might want
to look into is whether or not the functional approach that we generally use
makes sense, or if some other
approach, where the — you know, the fact that it is a virtual
asset in and of itself is what drives regulation versus what is
the underlying economic reality that
the virtual asset I don’t say that to think that
the virtual asset specific view is necessarily the right one,
but we should at least contemplate it.>>Thank you, and
maybe we’ll just work down the line. Ms. Udav we’ll start with
you and I see a >>Thank you so much again to
all the panelists really fantastic presentations. One question that I’d be
interested in getting some more clarity from
from the group is in relation to the question, Paul sort of
mentioned the existing volatility of the assets. In addition we’ve also seen a
number of events happening externally, for example the hack
of the Japanese exchange, I was in China,
cracked down on exchanges Korea cracked down on exchanges, we saw the DOC hack
as well as resetting that followed it. All these unfree dictate tbl
disruptive events thamd within the marketplace. So what extent
are the schaingz that are exchanging Bitcoin assets
and pertaining to other crip
cryptoassets how are they supposed to protect themselves
gaings the cost and potential volatility that’s going to be
exerted by these unpredibilityable,
seemingly uncontrollable events that are happening And to this question I think it
would be interesting to see how new this question is, and in that context to what
extent do exchanges need new tools, clearinghouses need new
tools or new way of thinking and/or modeling the kind of
risks that are coming up in this space. So that’s really one
question that I think would be an interesting one for the
working group going forward.>>I have two questions. One, Paul, you showed how many
forks taken place like just in one month alone. So I would be interested to
learn a little bit more on what causes
the forking, and do you have like any depth of insight into
how many times a particular Bitcoin could potentially And
then the other is much more for the working group, and that is
actually to have a better understanding for the broader market, if they have
exposure to like Bitcoin or Bitcoin futures, recognizing
that currently some of the offerings are commingled from a
fund perspective with existing contracts just is that like the safest way for
participants that may not directly be participating for
existing assets that he may be participating in, or should —
they may be participating in or should it ab segregated model
perhaps some sort of additional collateral that may need But that’s probably something
that the working group could look at more
holistically for these markets of these >>Let me talk briefly actually
a little bit about both of those questions. Bitcoin is a very unique
commodity for sure, it’s the only thing I can think of if there’s a fundamental
architecture flaw tore the computers break the underlying cryptography
overnight it’s a commodity that can go to zero and stay zero
forever. That’s an extremely — you can
burn for heat or food Bitcoin it’s
still in infancy so those risks have to
be considered. When whether it can be
segregated or not — in the clearing function. Forks can be limit sdples
anybody around here can create a slightly different version of
Bitcoin today. The real question ends up kind of like
circulating around whether it will have any value or not. So there are more high profile
forks than others just because people believe in the technical merit of those
four coins therefore exchanges those
other forks, monetize it but in
exchange there’s no way to commodity advertise it it’s a
complicated question because of the long nature of some of our
contracts you have to say do we believe this fork is going to be
supported and if so it’s going to have wide support with other spot exchanges, is it only going
to be available in a Chinese exchange that a lot 0 our
customers can’t get to, those are all west we have to think
about extremely carefully. It’s complicated but I liken it
to what a special dividend is for
stocks, when Microsoft issued their one time dividend a lot of futures were manual 33
adjusted the strike wise price to accommodate for that and I
think we’ll see something >>Mr. Stein was smart he heard my rule
he was we were going down to line so your card went up so
we’ll go to Steve and Mr.>>It seems there could be a good
opportunity to create somewhat of a framework around the
different types to have digital currencies if you will and even
looking, if you look at digital assets it’s nothing new to our
world, digital currencies can include things like points and
gaming tokens, cryptocurrencies as well and when you brick down
cryptocurrencies you can look at types that have exchange value
versus utility. And just understanding how those play
against something like a points program, where you can buy goods and
services at any airport now you can use your points to pay for
food and drinks, et cetera. So there could be an opportunity
to just kind of outline a framework
around how these cryptoassets fit into the umbrella of digital
assets that we all know and>> It’s actually a very, very
interesting obvious t. So we get asked a lot of times clients
ask us to speculate as to what is a particular cryptoasset. So we look at a number of
different characteristics today, we look obviously at what’s this cryptoasset
issued as part on R of an IC 0, was the I
cro CO before or after the dhul project was launched, what was
the purpose of the underlying technology for the coin. In fact, does the coin have
utility outside of the technology for which it’s
supporting, meaning is it used in currency. So we start looking at a number
of different characters, and I think it’s an excellent point,
and I think it’s something that the regulators are going to have
to look at going forward.It’s not just — the issue isn’t
whether it’s been issued as an ICO or not. As I said, I think you can
stretch the how sdwrae ie test way too far
to include everything if you want and I think it’s
meaningless. But there are criteria and it’s even
mathematics and probably quantitative. Keep in mind this
agency does have experience in this area. Today there are
futures contracts based on stock indices. And depending on
whether the stock index is considered broad based or narrow
based it falls basically under the jurisdiction exclusively of
the CFTC or shared jurisdiction with the SEC. And these inld
cease change. Everybody knows about the 21 A
Dow order but there was a 21 A order before that against an
international exchange because in fact the characteristic of
their index had changed and they weren’t keeping track of it and
it moved into a So there is precedent for coins
that might change characteristics, and for
monitoring it and recognizing it falls under a different
regulatory scheme.>>Mr.>>So despite all of the
excitement and growth and the numbers, Paul, you shared,
institutional involvement still remains quite low. And one of the key drivers there
is regulatory clarity.Obviously, what are these things? That’s
very important. Another one that we hear over
and over again is how do the AML, the
OFAC, rules affect this space. I highly encourage the committee
to Because without a solution
there, a lot of the infrastructure that supports the institutional investor isn’t
coming into the marketplace. I would point to the success of
a member organization like FINRA, and how
well self-policing hation kept up the standards so that anti-money
laundering rules and regulatory compliance
are in force, there may be an opportunity in this space to do
something similar. Thank you.>>Daniel Gorfine: Yes, please,
Mr. McHenry.>>Yes, is there any concern
about the concentration in the mining function? I understand
they’re responsible for maintaining block ledger I
understand there’s a great deal of concentration in terms of
mining amongst four or five large organizations, so I was
just wondering if that’s an area that needs to be regulated as
well, or what the >>So mining concentration is
certainly an issue. It’s something, though, that I
think naturally over time we’ll see
this concentration become diffused, for several reasons. One is that the in many one
input to mining is electricity, and so
miners are going to be seeking the cheapest sources, most efficient energy
sources they can. And so while we have seen that a
lot of mining activity happens in China, where there’s a lot of
excess hydro activity, for one thing we’ve seen an assist from
the Chinese government in essence pushing mining out of
the country, so we’re going to see it go But also we’re seeing a lot — a
race to find efficient sources of energy around the globe.
We’re seeing it in Iceland, we’re seeing it coming up in
Nevada. We’re seeing it in Washington state. And so I
think that’s going to drive I think also folks who are
building the technology fully recognize the
risk there, and are looking for technological >>The other thing is keep in
mind that the mining issue only really
exists in a proof of work blockchain so if
it’s a proof of stake where there’s different methodologies
of stideing who is going to conclude the block and get the
fees associated with that mining becomes — is not an issue.>>Okay, if there are no further
questions or comments, I would say based on this robust
discussion, and I think we’ve identified a number of worthy
work streams or maybe traps too many worthy work streams, I
would move that the TAC recommend to the commission that
it consider creating a subcommittee on virtual
currencies. Is there a second? Are there any questions or
further comments at this time? Okay, I will now call for the
vote on the motion. All those in favor of recommending to the
commission that the commission consider creating a subcommittee
on virtual currencies, please say aye. All those opposed, please say
nay. Any abstentions? With that, the
motion carries and remarkably at this time we’re going to be able
to keep with our agenda, which I didn’t think we were going to be
able to do time wise as we got started here but the TAC will be
taking a one hour lunch break, we will resume again at 1:30
p.m.. So thank you all. (Lunch break.>>Okay. We’d like to call the TAC
meeting back to order and begin our next session with a discussion of advances in
machine learning and artificial intelligence and their current impact on impact
on markets, market participants and regulators our presenters will
help us think about our role in markets, and will help set the
stage for next two panels which address automated
trading and cybersecurity. With that today our next
presenter is Tim Estes president and founder of Digital
Reasoning. I’ll turn it over to you, tym.>>Tim Estes: Thank
you very much. First I wanted to thank the CFTC
for the invitation co to come and present. It is a really remarkable set of
people here around the table. The visibility you have into one
of the fastest changing industries in the world is also quite
interesting to see how you dialogue about those things, how
you end up trying to drive the adoption of change. So what I came to talk about
today was really artificial intelligence and how it’s
applied in this industry, in an area that I think is going to be
very interesting to you particularly in terms of how conduct risk can be
managed in very, very different ways, than This is an area that clearly for
some time, and judging by the rate and frankly the volume and
the size of certain finds and activities, it feels like
technology for all of its innovations has not turned back
the tide yet. And I want to give some hope. So today what I hope to do is
tell you a story that is interesting and is true, so stories that are true in AI
are really important, you’ll find there aren’t as many as you
think. Without any great hype wave there is reality and there
is all these hopes.And we’re going to talk enough hopefully about where hope has turned into
real outcomes that will potentially make our market safer, will make our
population and the American people and investors around the world feel more
secure, with legitimate reason. So without further ado I’ll just
kind of jump into it. Some of this is a little bit of
the story of the company I was
really proud to build from college years 17 years ago, and so I’m a very proud UVA
grad and WAHU, and I’m very happy to I’m in Virginia, and live
actually in Tennessee, so the other person — group I’m very
proud to represent is Nashville. I think we have at least one
person here on this panel. So AI Nashville and Oahu, it’s
part of the personal story, and let’s jump into the more
interesting story about how So this is a very interesting
graphic that came my way, later about a
little over a year ago. Those that are in investment research
and finance might know what this is. This is the mentions and
earnings of artificial intelligence because it turns out emp company now is AI
company if you listened to earnings calls. So like every bank is a FinTech
company it looks like every finance company is looking to be
A AI company or both. What is AI actually being used
for and I won’t get into what artificial intelligence really
is because think it’s better explained how it’s applied
because the truth is artificial intelligence is just the way
software is going to be used in the future and the way it’s done
today. There will be no software that has no AI in it
within a decade. All of that will be commoditized. So imagine systems where the
code you write today may look different
and change without you or your IT infrastructure changing it.
And that being the norm. We have a transformation of
looking at this as engineering, and looking
at it more like biology of systems we can manage by
training or perturbing, but not because we can connect all these
dots anymore, they will be too complex. And so we’re ending a phase that
moves beyond kind of an early phase,
of computer science it’s going to be very different. Now, where
is this applied today? Most of the application of artificial
intelligence in the world is getting people to use more of
their attention to look at advertising. To spend time watching TV
online, to spend time checking your news feed, whether the news is true or
fake, and what’s interesting is that that’s because there’s a
heritage from the west coast and other places of using technology
that way. And I put this out there because
I want this to be a little bit of a call to arms, threerl the generation
I’ve grown up in and maybe the generation younger than me there are ways to use
this far more interesting, one of the ways we do this outside
our domain is we’re center pleased and honored to be
involved in taint human trafficking. We view that as one use case
where AI can do great things for good.And I think this is a case
in terms of monitoring our markets, which is equally
interesting. We started at Digital Reasoning
in a post 9/11 world of unlimited amount of chatter with
all kinds of risks in it, that people did not know what was
wrong until it was way too late. And that was a world that many
of us that grew up around that time, and I was getting out of college at that
time, sort of swore to never let it
happen again. I know many people here that are
based out of New York, you’re in banking, but many of you had
experiences, had direct knowledge of people and have
similar conviction And so that is something that was very,
Learning about language is going to come into play here, because
it turns out that human language, human
communication is potentially the most
interesting signal that any machine has ever had. Why is it
interesting? It’s actually used to educate every child we have. And if we want machines to be
more human like we have to educate them, and that means
they have to understand us they have to understand our language,
that is our great resource. Now taking that background to
Wall Street, the country obviously expempsed a very
interesting and similarly traumatic event, though not
violent but traumatic in a lot of other ways, with the crash in
2008, and all the consequences that came out of that. And the company we built,
Digital Reasoning, we saw that there would be need there as
well. So a few years later, two or
three years later, I will credit the financial institutions, some
of which are here in the room, with a vision to actually say,
could you use technology to understand human language, what
people are talking about, who they’re talking to, when they’re
talking, what are the behaviors that are involved in those conversations, what do they
show. In a world where people are
looking to apply artificial intelligence,
there basically was belief that all
the processes that were humor yented could start to be understood and used
for all kinds of new applications. And one of the
biggest ones was around So I will go into that. What we’re seeing now is broad adoption in financial services
on the concept of AI, immense belief
that it will transform this industry, and I want to talk very concretely against
all this hope, there still is, you know, kind of a giant suspicion of our
institutions that manifests sometimes in very
extreme ways, it has to be improved upon, it has to show
there has been a change. And when we see the data that
emerges from very recent scandals the last two to three years, we find that in
chats, in emails, in content, our
egregious behaviors, that if we had known
what people were talking about, we would have So the number one premise that
emerged from our work in Wall Street really starting about
2012 and 2013 with people like Goldman Sachs and others,
is if you want to predict human actions
you must first discover human intentions. This is a very unusual thing for
an action to happen before someone expresses an intention,
we generally call it a sociopath. For normal human beings that
actually just want to get richer and don’t care about the rules, they express
things that show that’s what they intend to do. If it’s a complicated situation,
such as price fixing and manipulation of markets, it
can’t be done by one person very often. Which is why we’ve had
so many collusion scandals and issues the last several years. So I think what we found was
that the same thing that could find
aliases of bad guys and different types of intelligence
to prevent them from attacking our troops abroad, or our people
here at home, that that technology could find very
subtle indicators of intent inside And that finding those
indicators was a way to preempt these things from So in 2012, we started working
with several of the leading banks in the world, such as Goldman Sachs,
UBS, and we have over the last several years actually
implemented some technologies that I’ll show that apply
teaching a machine about language, that apply an approach
to find certain kinds of conduct which previously no machine could see, and much less could a
machine process its scale at billions of emails a year. So we would not have known
frankly that was a place we could have been helpful and
served a more valuable purpose without the guidance of these
parties. So I want to start by saying some of our leading
institutions made this investment not as a direct
reaction to trying to comply or deal with
consent decrees and other types of activities; they knew they
could do better and they went out and searched and they found
it and they invested and they built this preemptively where
they could. And some of that work has been
broadly recognized, recognized in terms
of how we take essentially military oriented technology and
moved it forward. So let me get into a little bit
more of the meat of what this was. A lot of this was back drop
knowing which was a public meeting and giving a little bit
of the story. The biggest barrier of adoption
of artificial intelligence today is
the intent to process a machine from
data there are a myriad of technologies and techniques that can take well
formatted and data, that can predict
things and scale those models across a lot of data. The hard
part is we don’t have the curriculum to educate the
machine in nearly all cases, and the data we want to educate it
on is generally highly sensitive and highly protected.For good
reason. It’s customer information, it
has deep personal private information in it, so we have
had to figure out ways to accelerate the education of
machines to make this type of approach economically viable and
effective. When you read about large AI
failures, systems that cost 60 plus million dollars in multiple
years and don’t deliver something, you’ll see more of
these in the news. When you have this much excitement, if
you follow the gardener hype cycle, it is generally followed by a
trough of disillusionment they call it. And I have a hunch in 2018 we’ll
see a little bit more disillusionment than hope,
because how could we have more hope than hype than today. And
I’m going to give this advisory committee some ideas about what
we’ve seen and I don’t say it in the definitive way of this is
the case I can only state what we’ve seen and where we’ve seen
some effectiveness. But what we’ve seen is if you
make the investment in educating the machine not by sheer brute
force but by building a process that allows you to know where
the data is in an enterprise, in a financial services
institution, and allows a way to get the most yield from human
time, which is very precious, to teach a machine, you can show
much faster effectiveness of AI on almost any problem. So starting from that high
level, let me go a level deeper into what we have been doing. As we have tried to take this
automation approach of teaching a machine, and we have used
examples and unfortunately still one of the very best datasets available of a lot of
bad behaviors is Enron email. So those that have worked in
this space know the status very well it’s occasionally fun to
find something you never heard brought up before, but teaching
a machine how to figure out that there’s an attempt to conceal by
deleting reference toss very specific information, is a kind
of subtlety that previously wasn’t possible. And we know concealment,
secrecy, are generally leading indicators of other kinds of
behaviors which this commission actually has to run enforcement
actions against. So to put those pieces together,
the second — the first big piece that I wanted to make kind
of clear in terms of our judgment of what’s important in
this space is the technology to educate the machine, and being able to focus
on how you teach machines. Education of machines will
become an increasingly bigger issue. The next part is we’re getting
beyond what I call the transistor era
of AI, where you have little building blocks, individual
analytics, that do certain things, being very interested.
There are many options for most of these analytics, and just as
people rarely compete on transistors anymore, we’re moving up a stage, the CPU
of AI. Of integrated systems, much like a brain has very
different functions that have to work together and share
information, intellectual signals across it, to function
as a person, these AI systems have many different components.
They are designed with pieces that flow information one to the
other, pieces that deal with signals
coming in, context that’s accumulated over time. What we
will generally call global knowledge. The using of knowledge that’s
global to reinforce how you predict something that you’re
seeing again new. And it’s that local and global
that complete integrate systems becomes more and more
human-like. Because these systems, the secret of these
systems ultimately is they have infinitely scalable memory
compared to human beings. When this works the way it
should, whether it’s a year from now or five years from mow, I do not think
10 years. Almost nothing that has happened
in the past that was a problem should ever happen again, if we
have access to the information to feed the machine. Because it should see, with
perfect recall, what has happened in the past. And that is kind of — I’ll use
that to seed a question to this whole group later, of what is
important to teach a But what we have seen is that if
you look at this integrated system vision a critical part of
it is understanding human language. A lot of our
investment outside of how you teach a machine in general, is
what can you do with human language. And I’ll give you very direct
like kind of walk through. So when you’re dealing with
communications which are honestly probably the most interesting
part of determining what happened — because if you have
an email or chat where someone clearly said something, there is
very little debate after that. It’s not circumstances, it’s not
a behavioral pattern. And so we will get these kinds
of chats, that’s been where a lot of this bad behavior has
been for the last four or five years. Some email but more and
more in chat. And we’ll pull out the metadata, who talked to
who, when it happened, what the building blocks of words are
that are used. Beyond that, what are things
that are groups like electronic trading, what are phrases that
the system infers. What kinds of entities come up.
People, quotes, activities, monetary units. And then being able to take
whole regions of text based on
previously seen examples, and extrapolate that these words
together — not because of the specific word; the legacy
systems that are in place called lex con
basis that’s how they worked. Working with banking was taking
the symbols and having the systems figure out where none of
the words match. Which is what a human can do, a human knows
this word is similar to this word without having a big giant
dictionary in the sky. And so we have built up the
ability to find these indicators and show them in applications,
what we call intelligent systems and that’s very much like a
human reading these emails, reading these chats, and
triaging it to know what’s important. In systems today that have 99
percent false positive rates and consume a lot of resources, are
now moving into systems that have one 10th those false
positives, or recall rates of 75 percent on certain models, so
just revolutionizing this entire area, because the system can
figure out how to disqualify things that are false And now we’ve moved beyond that
to go not just against text we’re now
against audio communication, that’s the next frontier. To be
able to get accurate transcripts from traders talking, with
dialects, in trader-speak, because with a
true machine learning system, you can actually learn the language
model, if you will, the domain, the education, of a certain
area, and have the system So this is one of the last
pieces of puzzle is the ability of it to
learn from customers’ data is
ultimately the ability to empower every customer to take
the best practices they have, in terms of enforcement, teach a
machine, And that is very different. Now we’re no longer
going to be human resource limited with high index
for investigations, in theory those investigators can teach a
machine and what they see can be seen everywhere. As I mentioned earlier, we have
learned how to build certain models of
conduct using a process that defines, finds data, represents that
information for dicso and evaluates it and does things
like boasting, as you’ve seen in your own activities in your own
enforcement actions it is remarkable how people do
something terrible and they actually brag about it on Tom’s
later. Why? Because the systems in place weren’t
watching. So I’ve talked you through a
little bit of the process of how you would go through and use the
language of things like chat to find certain kinds of issues.
But that’s just the beginning, really. When you find an issue
in a chat, it only really makes sense if you can put it in
context of what a person has done over time. Because these things are really
a single event in terms of talking about something and
leading exactly to the market manipulation event. There is
normally a build-up. There is something to the effect of
interest in a deal, followed by secrecy, followed by changing the venue
to a an unmonitored communication, expend then
moving forward from there into the terrible activity to make
themselves money, and then to hide it.And whether it’s
volatility indices that are potentially attempted to be manipulated or it is FX, in
LIBORs we’ve seen in the past, the instrument may change, the
intentions of the human being do not. That’s just the latest thing
where the rules haven’t been written yet. So to be proven out against the
future is to understand intent, because humans don’t change. Now, clearly this group is right
at the forefront of trying to drive
change through enforcement. I thought I would bring up a very
recent one, and without isolating any particular group
or bank because there’s ones for a lot of people that have
happened in the last two or three years, I thought it was
interesting to look at the kind of language that it’s almost
hard to believe a computer could But the language of okay, but
what sort of level do you want to push it to. That, in the
context of the surrounding language, can actually trigger a
machine to say this needs to be looked at. And it can now. To prove it, we actually ran
some CFTC transcripts that were provided from cases, and we had to look at
traders that have been I guess — what is the old phrase?
Names have been changed to protect not the innocent in this
case.But this is real data. This is data that you’ve used in
your investigations. As you see, when people type in
chat, they don’t really care about getting the spelling
right. And when people talk in chat
they’ll use odd slang like whack it. But when you see a deal, and
then you see people acting in a certain way with some examples,
the system can flag certain events. And I don’t act like this is
easy, I don’t act like this is trivial, in fact there is a reason a 70-year-old
company that took 10 years in R&D to
even though investment banks to try, even
working with banks with their expertise to begin to teach a system to do this,
and I think that’s why you’ll see the disillusionment because
many people will promise as we like to say often imitated but
never duplicated, many people were promised, but there was
real hard work and if this was easy then I do believe banks would absolutely deploy it
and some are. But it’s not easy it’s because they’re willing to
put in advanced technologies to prevent future risk.So we have
run some of your chats and we found certain collusion and
boasting events with the models we have,
and we would hope that the kinds of data that we’re running today
live, in about half of the top 10 investment banks in the world, where this reasoning is
used in production today, that this is going to eventually
prevent some of the activity. Because as we understand human
intention and behavior through language, and through other types of data,
we can actually begin to preempt some of the behavior that has actually made
us all not very happy. So I did want to end by saying
we love certain partners in this process, we’ve worked closely
with partners like NASDAQ and being able to pull together
other data, market data, because language is a piece of the
puzzle. We argue it’s one of the bigger pieces because it
tells withdrew the smoking gun is. But it’s part of behavior.
And as we get into behavior you end up And so this is all converging to
something which is around what we call holistic surveillance,
and holistic surveillance from many different channels, text
and audio, and market behavior. And eventually, a behavioral perspective that sees every
individual over time, and looking for changes in that
individual’s behaviors, attitudes. It opens up the door
to many things. It opens up the door
particularly to avoiding the activities versus reacting to
them. So I’ll leave you with a couple
of questions. The first is now that we have
seen what is possible in some of these places and as they work to adopt it, we
do believe that the traditional approaches of using large lists
of words what are called lexicons, is a failed
legacy technology. We believe there are other
approaches that are also, you know, still failing us today. And I would — I guess my
curiosity is, how long, when innovation is
shown and proven out, is it reasonable
to wait to say it’s a new standard? A year, two years,
five years? How many things have to happen? So that I’d
love to have, because I think it has to be reasonable. It is not
free, it is not easy. But obviously, given the import of
this, that is a question I would love to And then a second question is
how — are there ways that we could start
to better enable, safeguard, the ability
of all of our members, people in this
industry, to share models so that they can learn from one
another. If educating the machine is the hardest part,
then you would expect, if we can create some levels of
exchange, of derived knowledge — not data, I want to be very
careful, I know there’s a whole other challenge in sharing data
— but there is a way to share
learnings, and operationalize them. What are ideas around
that area. Because we think once a few banks start to get some traction with
this and show the outcomes and approach that it’s things that
should be able to be shared and no bank, no firm should worry
about potential risk with them trying to do a good thing. So I leave you with those two
questions. I’m happy to answer any questions you have about
this. I haven’t touched on areas of revenue, areas of where profiling
customers, finding insides, market color,
there are so many applications today on this space, but I did
want to focus on one I think there’s plenty of definite. And
I guess at that point I’llened my opening remarks and just say
thank you for your time.>>Thank you very much tym. The goal of
your presentation was to be thought-provoking and that was
certainly that, and you start thinking about the application
of these technologies in the use case that you described, and
there’s many other potential applications or use cases of
this type of machine learning technology. So it certainly is
quite thought-provoking. But you posed some questions
which is great because it makes my job easier, so maybe I’ll
start with Charles, if you >>Maybe I’ll ask another
question instead of responding because I’m out of my depth here
on AI.Where do we think we are in terms object expectations
versus reality mismatch. We talk a lot about hype cycles
being, you talk about the period of disillusionment I think was
the expression you used where is AI now versus where people think
it is and how long lit take for AI to get to where people think
it is.>>I think about it in terms of what are the kinds of
use cases where it can be effective today. And then what are the promises
use cases where people are going to have to The number one use case today
where it can be effectively used is in triage. A sufficient level of judgment
about many kinds of data, to say this
data is not worth human attention, and this data is.But
I think there are all kinds of use cases, whether it’s in
sales, whether it’s in risk as we described. That is something I expect will
be ubiquitously adopted and could dramatically create
productivity gains Now, let me give you an example
of where I think maybe some expectations have been set and
maybe they won’t be met. 30 years ago, actually I would
even go back as far as 45, 50 years ago with the first wave,
there was a renaissance in the area called expert systems.
Where you could teach a system many, many, many things, and create
rules, and scenarios, and the system was
there, essentially as a collective knowledge of all the
people that taught it, to help I think for two or three
reasons, how hard it is to educate the machine, how hard it
is to back-test some of its judgments in certain context,
the lack of transparency in new deep learning models other kinds
of areas, it’s going to take some time before effective
expert systems in the new AI return the value that has been
promised. And so my belief is that triage
will be adopted very radically, but I think to be able to take
big bets on building giant expert systems with lots of
knowledge, and be sdploynd in what they really produce, we did
this already in health care and I think it’s really hard to
teach a doctor to be better doctor because what doctors
taught that system end up biasing that system tha,
people start rejecting it, don’t adopt it. So you have to think
about these human dynamics because human nature doesn’t
really change, and having something do the grunt work for
you, people love.Once they trust it. And the second part is it
telling you to do it differently than you’re doing it, that’s a
harder thing to pull off. So and then you enforce it from
the top down — if you know it works, let’s say you know your
expert system is better than most of your employees. You would force to tell your
employees, I don’t care about your jlt anymore.There’s going
to be things that society has to work out and I think that’s what
will hold back some of these things. And there will also be
a lot of things that we’ve commoditized that transistor
level of AI, but without systems you’re going to have a lot of
people try running whole programs on a set of transistors
barely wired together and it doesn’t really work on real
data. That will be the other side. If you don’t realize how to test
it on actually data versus test data
because a lot of big data in AI, the
recognition, cat detector, those have been done because we had
massively pro curd datasets where we could teach the machine
and those same models with different types of data prove
radically less effective. So the generalization of the
machines is still very much a work in practice. Which is why the most dwapsed
work at places like deep mind are
focused on very different techniques versus education
they’re on environment, to educate it but we’re still very
early in that actual science.>> We only have time for two more >>Thank you again tym for
presenting. And it’s good to hear that Goldman Sachs is a
client. I had a question, you spent a good amount of time on
surveillance, or conduct surveillance. I’m specifically responsible for
electronic trading with six businesses and in that context,
where you’re building automated systems that need to have very
discrete and reproducible behaviors, how do you foresee AI
influencing the sort of trading decisions made in those
businesses.>>Tim Estes: I think that’s
actually — so the easy answer is the humans that are making the decision, I can
see AI being kind of the chief filter ahead of the information
that drives that decision-making, and I think the
step after that is going to be how do you have all that
knowledge come together to where you can actually have a shared So I think essentially going
from radically augmenting the information that’s important to
make decisions, to essentially creating committee decisions where the machine is a partner,
and then looking at a lot of data and a lot of outcomes over
time, you’ll probably over a three to five year horizon, I
think not a one to two year, you’ll look at can we let the
machine do some of its Now, there’s a lot of risks in
that, like actual decisions based on data that are not just
sort of heuristic decisions. And I know this actually
happened in the market. So this is not — you know, science,
future. This is present.But the present is very dicey. So I
think that we have work to do to figure out how that would be
managed to regulate it, and it’s going to take sort of sandboxes
to get there. So thathose are my hunches those are We’re going to go from triage to
aton me over time, the question is how long and what are the
safeguards between those two end points.>>Let’s go with two questions I
will go with yeah sha.>>It’s something you’ve asked of us that is about how do companies
come together to help increase the
sharing ore enhance the intelligence of the surveillance
system. Like what are you thinking, is it like the phrases that our individual
teachings may come up with, help come up with a bri library of sorts?
What are you thinking in that context? Just so I have a
better idea.>>Tim Estes: Well I’m actually maybe three steps
behind you, in thinking that far in implementation t three
steps are, are there essentially safe
harbors that allow banks to experiment with changing
derivative artifacts from data. Where they don’t run —
basically they’re not — they’re lowering
their risk threshold to try. Because I actually think it’s
actually the legal and the concerns that hold it back.
It’s not actually technology. So I think if there was
encouragement to have the freedom to try, and
some safe harbors for that, you would see a natural sort of
adoption of the market to try to create network effects on key
issues.But I just think right now there’s just a lot of risk
aversion to the possible side effects. And if we could make
an impact there at the governmental level, then I think
the market will take care of the problem. Because it will be
naturally to share that risk knowledge for essentially — you
know, well intentioned self interest.>>Thank you, tym we actually
discovered>>It’s Amazing.>> Thank you so much Tim. I guess
some of the questions we had talked about in this panel
already are relating to new products, there
are new asset classes that have been
created look to cryptoassets. In the nascent new products how
do you deal with AI when there is really no data, or the data
is being created the history of the data is very new and it’s
hard to discover what filter mechanisms and modeling you
needed to make this data. Particularly given the
marketplace that we live in, sourcing that
data using very special syntax form is probably very expensive.
So how do you kind of overcome that issue. One interesting sort of point
that you made which is fascinating, which is we want to
make machines human like in certain respects and I think
that would be important the point I was saying about
surveillance, looking at language, but what about when it
comes to looking at traineding, maybe we want to get rid of some
humanity in relation to that. How do you make that
determination between what parts of human beings we want, and
what parts of machine we want at a certain point.>>Tim Estes: I
should think the questions are very related. So to go to the
first point, the strategy we’ve taken ultimately
was about finding human behavior by finding intentions that run
ahead of behavior. And that is iment of products. So we think that if you look for
human risk, whatever the current thing is, even if there’s not
much data on that, there’s still plenty of data about people
doing bad things with other products.So I think we have a
little bit of insulation there, I don’t necessarily say we have
full, I actually think it makes more structured data or metadata
driven decisions much more technical because those require data of
that kind and fit doesn’t exist it becomes a problem. So I think there are good
proxies for new data by looking at intention. For me that’s a
great case for human being implemented in AI that we’d want
to model, chtion find the parts that are invariant and focus on
getting those first. And then work on the more subtle areas. So I think that’s a strategy,
that’s a human-like principle. Let’s take an example of a
human-like principle that we don’t want, we want to avoid as
much as possible.Prejudice, and bias, and mistakes. So in those areas, we want the
objectivity AI can bring and the transparency of it. So that’s where I think that you
actually have an ability, so if you think — if AI is essentially
just a label for educating machines, transferring knowledge
that’s — knowledge that’s in data, and in human heads, and
judgments against in a data and moving it into a model or set of
models, the giant family models — right now I think we talk
about models in terms of a few things, a few patterns there will be thousands to
millions of little models like little brains running through
our industries not too long from now. And that’s why it will
become impossible for us to really look at it like a code in
an engineering problem it will be like a bunch of cells and
knowing pha kinds of things to perturb the system with to make
the cells do certain things. So I think engineers will become
more like doctors.Having said that, in this case what you want
is you want still some of those engineering properties to have
audit and reporting — I mean, we can’t really dump our brains
out on to a table, we wouldn’t know what’s in the stuff in
here. We have to have it write out a lot of like you give it a
test, right? How do you explain what somebody knows. You test
them. But what if you have nearly
infinite processing and storage, you could run a test which is
almost entirely exhaustive. So something that a human could
not actually sustain, to go deep into all the things they know,
and why they did something. You can have a machine just keep
going and going. So it becomes a lot like
inductive proof in math. And I think that there is sort of a
next way of looking at this, which essentially is more in the
inductive space, but it’s such scale that it And that will be where it’s kind
of between human, where it’s not fully certain, because it’s not
just deductive, it is learning, it’s adapting, but it’s at such
ridiculous inductive scale that you know it’s essentially true,
as much as we know many things are true. So that’s my guess.>>Daniel Gorfine: Thank you
very much tym we appreciate it. And this is actually a very good
segue to our next panel topic, so I would now like to turn to
the next topic on our agenda, in which our panelists will discuss
developments, challenges and risks around automated and
algorithmic change technologies and potential areas for
regulatory consideration. Our panelists are Larry Tabb of
TABB Group, Bryan Durkin of CME and
Yesha Yadav, professor of law at
Vanderbilt university. We’re going to start with Larry and
rely on a very, very advanced technology called the telephone.>>And the presentation that
he’ll walk us through.Larry I hope you’re there.>>Larry Tabb:
I am here, are you guys there, can you hear me? Can you hear
me? Hello?>>Daniel Gorfine: I’m told
Larry is on.>>Larry Tabb: I am unmuted,
can you hear me?>>Yes we can excellent.>>Larry Tabb: Okay
great, thanks. I want to thank the
commissioners, the chairman, my fellow TAC members. I’m going to kind of do an algo
training level-set for you. The agenda you know that I’m going
to go through pretty quickly on the next page are type of algos,
kind of the difference between future and equities, where speed
fits in, a little bit about method, too, but not too much.
And a little on algo regulation. The next page, slide 3, it says
three types of algos across the top. The way I look at the
market, even though it’s way more complicated than this,
there are three general types of algos, the first is alpha
generating algos, it’s about 54 percent of total equity trading volumes, a lot of
it is market making liquidity provisioning. The other side of
liquidity provisioning is basically finding opportunities,
and picking off liquidity providers. And then there are a lot of
arbitrage algos used to keep futures and cash Options cash, you know, ETFs and
futures in cash, and basically keeping everything in line,
price-wise. On the other side, the next set,
you have quawnd Taif strategies. They tend to be more technical. You know, very much data-driven. We think this is about 16
percent of total equity volumes of that, we think about 5
percent we think would be categorized as HFT but 11
percent is And then there are executional
goes there are buy-side algos, you know, where we view that
about 45 percent of buy-side flow directly goes through
algos. And then a lot of the trading
that goes through high touch brokers gets put into algos after that, so it’s a
very large percentage of buy-side, winds up going through some sort of algo,
again on the equity side. If you go to the next page, just
kind of a chart of the difference between equities and
futures. Generally, on the futures side,
the bottom right, products generally trade That they trade on an exchange,
there’s one queue, it’s kind of hard to do things over the
counter. And the queue is, you know,
first come On the equity side there’s a
tremendous difference. There are 12 exchanges, soon to be 13. There are roughly 33 equity ATFs
there are roughly 35 brokers and market-makers providing quotes
to institutional investors, there are 8 major wholesalers that are providing retail
execution, and four ping networks which is bairkt lay an
electronic market-maker. So from an equities perspective
there are roughly 92 routing decisions. Where do I go and how do I trade
and how doch of these different venues operate. So it’s a much
more complicated, and to certain extent there’s 45
formal queues to be first in line especially on the equity side with order
protection becoming first is very important
to how you get executed. So the next page, slide 5. When liquidity is fragmented,
the tick size is really important. Basically fragmentation and tick
size basically means that speed becomes really, really important
because there’s a latter of arbitrage between all these
different venues. And where to become — you know,
the top O of queue where up can get queue positioning becomes
very important and depending on your strategy you may want to be
at the top or you may want to be in the middle or there’s a lot
of jockeying around. So small size, and being very
quick becomes really, really important. To a certain extent
none of these venues have different charging mechanisms so there’s a big gain
in’ not gain-gain but in theory as to where do you want to be in
the queue. And not just a queue in one, you
know, platform, but the queue across the holistic market. And that has a lot to do with
whether you’re lit or dark, what the pricing model is, and how
aggressive do you want to be in terms of posting or liquidity
and how visible do you want to be. As I said, when there are
multiple venues or queues, speed becomes really, really
important. That’s not as much so in the
future world, where there’s fewer queues because to a
certain extent, if you want to be top of queue in a futures
market, you’ve got to get to the front and then size becomes —
you either — you know, the larger order basically stays
there longer until it gets filled, or
the person wants to jump that queue needs to pay a much higher tick size to
get in front. So in the equity side, speed
becomes incredibly important. Speed becomes important in
futures basically to tie cash and
futures If you go to slide six, also
depending upon your architecture, and what you’re
trying to accomplish — or depending on what you’re trying
to accomplish, architecture means a lot. So on the left
side of this picture you have a typical buy side
trader, or even, you know, a buy side executional go, trying to
determine where do I want to be to get the price that’s being
displayed, so I have the correct market data from all the venues,
put them all in order, determine where do I want to route. Whereas if you’re in an alpha
algo architecture basically a market-maker, a lot of times
they’re colocated at all these venues, and you’re dealing — there’s a very small time delay,
you know, just reading one venue
reacting to that venue, providing liquidity and taking
liquidity. So generally, market-maker algos Slide seven really talks about
method, talks about unbundling. As really, you know — and then
you’re looking at because of my
research gets unbundled I have to pay for it separately,
enables the buy side trade tor really take more control over
their execution process. Within method two there’s a very
significant change, especially in the equity market structure, price,
band dark pulls, dark trading, requires a lot of buy side end
of the day reporting and then we wind up with exchange driven
clearing, being much more open, but that gets postponed to 2020.
The issue here I want to make is really around market structure
regulation, and really the third major bullet. You know, because — you know,
fist ma wanted to push more liquidity into lit pools, they
tried to — they banned broker dark pools and tried to
limit dark NTFs but they opened up a channel internalize which
is going to pretty significantly change the use of capital, how
systematic internalized market-makers interact with the
buy side. And it’s going to pretty impact
how the European market center acts. And I use this to kind of talk
about regulation, which is kind of the next slide. You’ll see a little logo there. And if you go to slide 9, where
I have kind of my thoughts on regulation. And here I’m kind of probably
going to upset a bunch of people, but my thoughts on
regulation are that, you know, I think the regulators you should
be trying to create a level playing field not pick winners or losers or
program models focus on transparency
fairness risk, clearing, make sure the
margin is clear, clearing margins safety things like that
the other thing the regulators I don’t think focused enough on
is transparency. Increasingly as the market moves
become more automated the issue of time stamps and clock sync becomes
increasingly important. Where instead of, you know,
markets or regulators determining how we want our
market to work and how we think if we do these changes the
market will react, it becomes — if you look at where the
regulators are kind of forced the market to move, very often that’s not
what exactly happens. You have methods trying to close
down dark polls but turns on this
internalizeer process which that’s another whole other
discussion. You’ve got the implementation of
SEFs in the U.S. trying to create limit order books and not
seeing a whole lot of use of those liment real time order
limit books one of regulators market — a long
time ago, you’ve seen in the options market just overwhelming amounts of
data and challenges and quoting — you know, 600,000 instruments
on a real time basis over 14 exchanges. You wind up with some very
unintened consequences. Which actually, this is my last
slide — my bottom line on markets regulations let’s be
able to measure stuff. If we can measure, get down to
find enough grained time stamps, if we can get clock sync right,
we can really better understand how my broker is serving me, and
whether they’re actually doing a good job. And if they aren’t,
well then, you know, I can switch brokers. Whereas if I find up with a
sweeping market structure changes, very often they’re not
going to really turn out the way we wanted them to turn out. And so that’s kind of the end of
my two cents of pontificating. So like the carpenter said, you
know, the carpenter adage goes,
measure twice, cut once. I think our topic should be
measure twice, regulate once. Let’s make sure that we know
what we’re — you know, where we’re going before we wind up,
you know, And with that I’ll put myself on
mute, or be on the call. Thanks.>>Daniel Gorfine: Thank
you. Thank you Larry. I think actually on the list
here we have Mr. Durkin next.>>Bryan Durkin: Thank you very
much, first of all thank you Commissioner Quintenz for overseeing and
revitalizing the technology advisory committee. Chairman
Giancarlo thank you very much for your support on the
importance of this, Mr. Gorfine, very excited about your
leadership in this whole program. Commissioner Behnam
thank you very I say that because I have an
over seven year history, at least, with the technology advisory committee,
and the And if you go back and look at
my comments over those years, they’ve been very consistent, in the context
of applauding the CFTC for its leadership in this regard. And
in this evolution of these markets, and the automation of
trading. Because you have been leaders in
this regard in bringing this community together, the esteemed
colleagues, to help inform policies,
guidelines, and principles has been very formidable in what I believe to be the
leadership that the futures industry represents. The efforts that you all and we
have all undertaken together over this
journey have led to some very important and highly notable
principles-based guidance, involving a wide range of subjects that directly impact
and are related to the advancements in technology, and the progression
to an increasingly automated environment in which we all
operate. Among this, this guidance has
informed our industry, it’s informed the
commission. I’ll go so far to say it’s
informed the globe. In the context of risk
management, pre-and post-trade protocols, systems safeguards,
access to colocation facilities, messaging policies,
and no, Now, it’s not at all surprising
that traders today have increasingly
turned to automation to optimize trade execution, and to increase
their operational eftion sea, and
enhance risk management. With the development of automated trading, algorithmic trading has unquestionably emerged,
contributing to significant volume and growth across all asset classes, and providing
greater liquidity, tighter bid-ask
spryds. This liquidity generated by the traders in these markets in turn
is relied heavily upon all types of market participants to
achieve their risk management and their investment objectives,
and allows them to do so at a lower cost. Very important to recognize that
algorithmic trading, like nonautomated trading, engages in a variety of activities such as
market-making, arbitrage, hedging, and employs
many diverse strategies in each of these contexts to achieve
their risk management objectives. A significant proportion of
algorithmic traders active on CME group
markets, they contribute substantial liquidity
uundoubtedly, by providing continuous A major benefit of proprietary
trading that I would like to remind all of us of, or as has been referred to,
professional trading firms, was highly demonstrated and very
publicly demonstrated in the October
15th, 2014 U.S. treasury market flash rally. Where proprietary trading firms
played a key role in liquidity
provisioning, and price discovery, during a
highly significant and highly volatile market condition. During the period leading up to
and including most of that volatile period of that day, proprietary
traders, or commonly referred to PTFs,
increased their trading activity in the 10
years note futures market they proit provided math of the order book liquidity
at a type that ask spread, in contrast that day during the most volatile
period, dealers big dealdal dealers
widened their big-ask spread and pulled back their participation.
It’s important to note while the event itself was highly
significant, continued pricing spurred on by prop
traders or PTFs filling that void that was left by other
participants, allowed the general market population to
continue transacting even during the most volatile period of that
day, leading to a much more orderly marketplace. Now, the criticality of
proprietary traders extended liquidity during this important
market event should not be lost. And it’s something that I would
argue that this committee should take a look at going forward,
for the following reasons. As we evaluate the impacts of
new regulations on these markets, regulations must be
appropriately tailored to our markets. Capital treatment for
proprietary traders, for example, who make
markets in exchange traded derivatives, ignores the actual
risks and correlations between the positions of participants
offering liquidities on both sides of the market. More specifically, they lack the recognition to have delta
adjustments for options and recognition of netting sets for
options positions, which ultimately results in a
reduction in liquidity during stress conditions. This happens by applying
ininappropriate cost to market-maker exposures, and this is a direct result of the
supplemental leverage ratio. We recently observed the impacts
of this treatment in the equity derivatives markets, where spreads on the s S&P options on the fifth and
sixth widened significantly more during similarly stressed
markets in the past. Based on our market intelligence
we received we understand the widening of the bid ask spreads
to be driven primarily by capital costs again associated
with the leverage ratio. You now what is our role as CME
group or exchange in automated trading? Well I assure you that the CME
group shares, as I’m sure all of our colleagues around this table
shared, a common objective of promoting transparency, and integrity, in
our In doing so, in a manner that
preserves the vibrancy, the competitiveness, and Unquestionably, market integrity
is one of the cornerstones of our existence. I assure you it is the
cornerstone of the existence of CME group’s business model, and our company employees
substantial human resources, technological capital and
capabilities, to protect and continually
enhance the reliability of our markets. And also, to mitigate the
potential for market disruptions, through the usage of its risk controls in
the system safeguards thieve developed over the years. Our protections against spoofing
and other market abuses or integral
to these efforts. We view that CME Group has been
a leader in promoting integrity,
efficiency and transparency of global financial markets. We’ve
done so deeply with the guidance of the commission and the work
that’s been done with the technology advisory committee
over these last seven years.We appreciate the importance of
ensuring that risk management and
regulatory frameworks keep pace. Keep pace with the rapid
technological advancements thab that have
characterized the evolution of our markets in recent years. We’ve developed a wide array of
capabilities to manage risk and volatility, and to mitigate
market disruptions. And these are applied at all
levels of market participants. All levels of market
participants. These protocols include globe ex
credit control, price spanning, price
order quantities, messaging controls, stop logic, functionality, such as
circuit breakers, price protection points and kill
switches. Now, each of these is outlined
in some form or fashion underscore
guidance and principles that have been the good work of this
group over these years. And to drill into these a little
bit more deeply, just so we have an preerks of how far this
industry has come, when we speak of globe X credit
controls these are preexecution risk controls that are provided
and they enable our clearing firms to set credit limits for
every executing firm. It’s a requirement that our clearing
firms utilize these controls. Not an option. Our credit controls, which every
clearing firm utilizes, includes mechanisms such as order
blocking, order cancellations, automated email notifications,
and these can be set at We also employ a tool called
cancel on disconnect, that will cancel all resting orders in a book for a market
participant that happens to get disconnected from our system. Price banding, the supplies to
all of our products, all our orders are subject to price
verification levels, bids and prices well love and offers well
below outside the market the contract span and they’re
automatically rejected by our system. Maximum order quawngts. Every product has predefined
maximum quantity per order. This step ensures that the order
is not exceeding this limit. If the maximum quantity is
exceeded the order is rejected by the system.Messaging
controls. These controls limit the rate at
which firms may submit mass quotes,
and can block orders from entering the system. If the
volume thresholds or the order quantities are exceeded.Stop
logic functionality. Stop logic can automatically halt the
market for a predetermined period of time, in order to help
prevent extreme market volatility and price deviations. When it was triggered on May 6th
of 2010, that infamous day, stop
logic was the mechanism that reversed the course By halting the market for enough
time for liquidity to be replenished. Circuit breakers.
In our equity indexes and energy product, circuit breakers halt
trading for a period of time, when a specified level is
reached. In addition, the utilization of
daily price limits prevents trading at prices higher than or
lower than limits that are preset by our company.Price
protection points. Protection points act as a
control against excessive price swings
in a liquid market. These prevent stop orders from being
filled at significancely aberrant prices because of the
ab accepts of sufficient liquidity in that market.Kill
switches. Designed to allow clearing firms
a one stop shutdown of our CME globe
activity at the most granular level of a market participant. Additionally, more recently
we’ve just introduced inline credit controls that allow for
account-based limits that can be set at a highly granular product
level. Taking a look at more recent
activity, I just wanted to end by referencing how these markets are continuing to
grow and evolve, and having capabilities such as what we’ve
outlined, capabilities that have been the predicate of the good
work of this committee, has enabled
these CME Group’s markets this past
week, has investors turned to these markets to manage and to hedge the largest
cash market drop since 2011.In all instances, our market worked
as designed. These protocols that I outlined worked as
designed. Our risk mitigation tools and processes were utilized
extensively throughout the week. They worked as designed. Our systems performed well with
large volumes and open interest. Total open interest records were
achieved last week of 129.5 million contracts on February
6th. New daily volume records for
options of 9.2 million contracts on February 6th. Last record in that regard was
the U.S. presidential election. Equities and futures options
reached several records during the week of February 6th at 10.7
million contracts. The innovation, the advancement,
the engagement of this commission
and the technology advisory committee has played a very
formidable role in the leadership of our U.S. futures
markets. And having the protocols in
place, and the system safeguards in place,
to keep pace with the innovation of technology and automated
trading. And so I feel very strongly
about where we are today, and the opportunities associated with how this
committee can continue to help us evolve and keep pace with
that change. Thank you.>>Yesha Yadav: So my sincere
thank you to Vanderbilt own chairman
Giancarlo of course, chairman Quintenz,
Commissioner Behnam and of course Gorfine to
have the chance to serve on this committee really a tremendous
honor to have this at this time amongst such terrific
colleagues. In addition of course, the
incredible staff here at CFTC for all their hard work and
thoughtfulness in putting this wonderful program together it
really is a fabulously cohesive and thought-provoking program
that I think will give us incredible volume of work to get
on with, in the time going forward. So giving the incredible
activity that we’ve had, over these past few weeks in the
partnership. It is actually a good time to
step back, to stop looking at the ticker tape every single
minute, and to stop doing our — start doing our breathing
exercises and to have a chance to actually go back in time.And
I thought I might take you guys back to what seems like
prehistory now which is August the fourth 2012. And that was another weird and
problematic day on Wall Street as many of you will remember. And that was a day on which
knight capital an AFT market firm
experienced what can only be described as a bad technical
glitch. Instead of sending out 212
orders, test orders, into the New York exchange, a failure of knight capital to
update the software system on its router caused the system to unleash several
million orders into the marketplace, resulting in in
Knight capital accumulating losses of around $450 million, in just 45
minutes time. For a firm that only had around
$360 million in cash and other assets. And these weird an om lus costly
and disruptive events seem to be growing in We can’t forget obviously, as
Bryan mentioned the infamy of the flash crash, when a trader sitting in London,
in his parents’ basement, managed to contribute to events
that eventually led to the crashing of the U.S. stock
market for a period of time. In addition and perhaps more
mundanely we have seen flash crashes in the trading of individual stocks,
individual securities in the futures
markets, sudden disappearances of
liquidity that are all explained costly, and create anxiety amongst market
participants as Ho to how market robustness is supposed to look
like in our leading markets. And we thought it would be
interesting for our committee and for market participants
regulators and policymakers to think deeply about how we might
lever our legal system in ways that can encourage and deter bad
and disruptive behavior from taking place in
the first place, as well as for the legal system to be utilized
in ways that can create mechanisms that can help make
participants whole, to compensate participants in the
event that these disruptions end up happening. Now, one concern that has been
highlighted in the research that I’ve been doing is that the legal
system is increasingly unfit and poorly adapted to dealing with
the risk created by high speed, highly automated trading in a
very interconnected market. As many of you know if you’re
lawyers or if you watch the good wife and/or law and order, the
legal system is basically underpinned by three very
fundamental standards of liability. That underpin much of the
rulemaking that happened within these van
— walls as well as across agency across
ds. When we think about strick
liability it makes you liable irrespective of
fault, you are liable for all losses as a result of your bad
or disruptive conduct. And this kind of liability is designed to deter and punish extremely
dangerous, disruptive kinds of activities
that can create large costs and large disruptions in society. Then we have the workhorse of
the legal system, and that is negligence-based liability. Negligence-based liability
punishes unreasonable behavior, and makes you liable for the
foreseeable consequence of that behavior. And when we’re dealing with
negligence, that’s really one of the
linchpins of our regulations and the reasonable standard, the
reasonableness standard undergirds much of rulemaking
today across regulation in the securities marketplace. And then of course we have our
most serious standard of liability, which is intent. Intent or highly grossly
negligent conduct that is not in cases of fraud manipulation that
is really regarded as being the most expressively bad kind of
conduct that we can perform, and that is
punished by an intent based standard of liability. All of these core fundamental
guiding standards of liability are increasingly poorly adapted
to the marketplace in which we live in, which is governed by
high speed, highly automated algos, in a market that is
deeply interconnected, where the seriousness of the harm can seem to be far in excess of the
actual nature of the act itself. So when we look at algos, in
order to be able to transact in
milliseconds, in microseconds, we need to make sure that
algorithms are preprogrammed in advance of trading. Because as
Charles mentioned earlier today, we’re too distracted,
we’re too dumb to be able to follow along in real time. And
as a result, of course, we need to make sure that our algorithms
are preprogrammed in advance of trading to be able to deal with
the permutation, the different events, the different dynamics that will happen in an
ever — evolving marketplace with changing prices.And
therefore, the fact that we have the pre-program algorithms, the
fact that we have a predictive dynamic that
underlies much of trading today, whereal
goes essentially have to trans tact essentially in real time in
accordance with their programming Medes means in a
systematic, random error is almost That algorithms can easily MIS
fire, algorithms can easily fail to react to information the way
that we might expect them to. Algorithms may fail to perform
as programmed in circumstances that are unusual, because
essentially they’re constrained by their programming, and we
cannot possibly expect traders to take into account every
single eventuality that may possibly arise in that program. So we have preprogrammed
algorithms that create a dynamic where by systematic In addition, as we all know, as
we all know today’s markets are incredibly interconnected, as
Larry mentioned in his presentation, we have venues
both in the space which raibl to
synchronize very rapidly to new information that comes in. So as fine scholars have
repeatedly shown, markets today are incred bleap efficient at
reflecting new information at the prices at which securities
are trading. In addition of course that means that errors, miss fires and
disruption can also potentially amplify and cascade as they move
through the system of venues, such that an original bad act,
an original harmful event, can seem far, far more serious as it’s
amplified rapidly across the marketplace, in the prices and
the dynamics that exchanges and other venues have to deal with,
when they’re dealing with cascading events in And these two dynamics make it
very hard for the legal systems core
standards of liability to work as intended. If we think about strict
liability, for example, strict liability,
where you’re liable irrespective of fault is very difficult to
apply in a market in which systematic random error is
endemic. We cannot possibly expect, we
cannot possibly imagine, that traders will be able to build
systems capable of safeguarding against every possible
eventuality or circumstance in an In addition, the negligence
standards too are very core of our regulatory framework, is
straining. When we think about actions like
nightcomrrks or even the flash crash, actions that seem remarkably
innocuous, are remarkably expected in certain circumstances, can cause
incredibly costly harm to the marketplace as a whole.And as
we’ve sort of thought about the discussions throughout the day
today, the increasing
electronicification of our marketplace the increasing
automation of our marketplace means that we might see an increase in
these kinds of glitches and events happening such that it
gives us some room to think about how we should recalibrate, rethink, reimagine, reconfigure,
the reasonableness standard, to deal with the kinds of harms
that we are face nothing today’s marketplace. If we think about our own lives, control, alt-delete are my
favorite keys on the keyboard these kind of glitches happen
all too often. So how do we reconfigure, think,
again with a reasonableness standard when we’re in an
environment of a highly automated, highly electronic and
a marketplace in which preprogrammed, high-speed
algorithms are becoming the norm. Now, finally, and perhaps moms
interestingly from a philosophical perf, particularly
given tym’s earlier presentation, very insightful presentation earlier, how do we
think about intent in this market? When we look at automated
trading, we look at machine learning
algorithms, that are effectively designed to reprogram themselves
in response to new incoming information, what do we
do about intent? Behavior that to us would seem
bad, undesirable, intentionally malicious may to certain
algorithms be considered as being profit-generating, as
being effective, as being a useful
arsenal in their trading technology. And when algos are trading with
other algos, how do we capture our
vision, our understanding of harm,
intentional, and bad behavior, within the programming of highly
automated machine learning algorithms? Now, one big gap, given these
gaps nar left in the law, the cost of these gaps are going to be left to be borne
by market participants, exchanges that have to redouble their monitoring
efforts, or those who have to pick up the costs when firms do
not have the resources, when their actions are too serious
and too costly relative to their conduct, to make whole those who
are in the marketplace.And so someone has to pick up the tab. And that tab is either other
market participants and/or it is investors who have to discount
the capital they put into the market, to reflect the risk of More broadly, obviously, seen
from the perspective of this agency that has done so much to
lead international rulemaking in financial regulation, this is
an area which remains underexplored, underthought, and in which
guidance from us, guidance from our market can really inform and
lead the public dialogue, both publicly as well as privately
and thinking about and calibrating new ways in which to
bring the law up to speed with the pace of technology that is
currently underpinning trading and securities markets today.>>Daniel Gorfine: Okay, I want
to thank our panel. And with that, we will open it
up to questions, thoughts, or reactions to what you heard from
the panel. And again, I’d like to keep in mind the idea of
potential work streams that Let’s begin with Chuck.>>Hi, I wanted to thank the
commission and also the panel for this whole event. But these
really resonated with me, although I’m not really working
in this space right now. Most of my career was spent doing
high frequency trading, at big banks and hedge funds. And it’s fantastic to see the
efforts that CME group has put into automated tools and kill switches and
things like that. And I would just like to say
I’ve always been appalled that most of the larger organizations
I’ve been at, how few controls there were or how much — how
little transparency there was, even within the organization, in
the face of different kinds of failure scenarios. So I mean
this is not unique to trading, I mean, this is a general
technology problem, this is true in any kind of system with massive
concurrency and distributed elements, and you know the same
thing applies to some of the blockchain things we were
discussing earlier. And one of the major
organizations I was at, when something were to fail, you
know, could be a hardware failure, could be a software
failure, it would take hours or days to resolve the situation
and figure out what had gone wrong, until we eventually
completely automated things so that, you know, if a switch
failed in the data center, the head trader would get a dialogue on
list desk saying the options connections
to the exchange are down or something like that, and this
was a question that I was going to ask in the AI session, having built a lot of artificial
intelligence or recognition of pattern based
algos, one of the things that a lot of the newer technologies
especially deep learning neural network technologies lack, that
were present in a lot of earlier technologies, some of the
earlier technologies, was this notion of explainability. When something goes wrong,
because the algorithm is adapting, or because, you know, a new situation occurs,
there’s no explainability, there’s no way to understand why
did it make that decision. And because at a certain point, it’s
just numbers, it’s all these — you know, linear equations and
you’re trying to figure out — well, we trained it. Here are
some coefficients. They don’t explain themselves, they’re just doing dot products at very
high speed.I think it’s something that we should hold people to a higher standard
to.>>Okay, Mr.>>Thank you, I wanted to ask a
question of Larry Tabb if he’s still
available on the phone, in particular his
slide 6. Where he references that
executional goes are typically slower than proprietary algos. And then has the added feature
of it being kind of hard wired into the system.>>Is that different in futures
markets, where there’s more of a single
queue than a multi queue dynamic. And then second, does that
really result in a form of structural
disadvantage for parties using executional goes,
maybe disproportionately customers, in the sense of a small difference
making a big difference in terms of actual realized results in
trading. Even in a system where you’ve
had a prioritized fair process over fair result, and kind of the way
method is challenging now I would suggest there’s even a
fair process concern if those types of differences exist in
terms of tracking between execution and the proprietary
algos.>>Larry Tabb: Yeah, I’m still on, caw can you hear me?>>Larry if you’re there and
you’re able to hear the question, please go ahead and
jump in.>>Larry Tabb: I am here, can you hear me? Can you
hear my guys? Hello hello hello.>>We can hear you well.>>Larry
Tabb: Okay good. In terms of futures customer
orders versus equities customer orders,
that problem isn’t as big a deal, but depending upon where
the execution algorithm is, and how the data is structured and read, it could —
you know, it could be impactful. But not generally as impactful
as on the equity side where there are so many Now, the question is, is this a
good thing or a bad thing. Or does this — you know, create
poor outcomes for customer orders. That’s hard — it’s harder to
say, because it’s really difficult,
we had a business for awhile, before it spun off, of really
measuring equity execution, and just getting the normalized
information, even down to one millisecond, was very difficult. And actually, to really get an
accurate representation, you really needed to get into the
microseconds. And maybe 10 microseconds or
sometimes even single second — single
digit microseconds. And none of the brokers really
had the ability or I’m not sure we even had the ability to measure or, you know,
monitor On the other hand, the customer
orders have another advantage that the market-makers do not
have. Is that they know the full size. And so — to a certain extent,
knowing exactly what you’re trying to accomplish, that this
is really only 200 shares of — you know, half a million share order, actually the buy
side has tremendously more information than the
market-makers do. So 0 as well as the symbiotic
relationship, in that the market-makers are providing liquidity, if they
do not get fair outcomes then they’ll widen their quotes. So it really is kind of a cat
and mouse game. That said, what I believe and
what I said in my presentation that’s really important is just
that much better mechanism to measure and
benchmark execution speeds. Because then you’ll have a
better understanding of whether you’re being treated fairly or
if there’s a problem that we should be worried about. And so unfortunately it’s really difficult to answer that last
question, but I believe that both sides have advantages in different ways,
and it is a symbiotic relationship between the liquidity provider and the
taker. And if the takers have an
advantage, the market-makers would go out of business, and
the market-makers had a super advantage, then the
investors would clearly see harm in terms of getting executed. But that said, I believe better
— more fine-grained time stamping and
better computer sync would help that. Now, I would like to provide —
ask — provide a comment myself upon on the last presentation
from the professor from Yale. I think we’ve come a long wairks
you know, I don’t think we should always be worried abouter rent or aberrant algos
and their impact but I think the exchanges and regulatories have
done a really good job in terms of implementing kill
switches and limits and the circuit breakers and things like that, to stop the
market from poor reaction. Now, does that impact the legal
system? No, but I think it limits some
of the damage that occurs.>>Daniel Gorfine: Okay thank
you. We’ll take two more questions or comments. Mr. Hey Myer then Mr.>>Island I’d also like to start
by chiming in thank you all for the forum and all the work that
went into canceling and then rescheduling thank you very much
the presentations today have been terrific. And this is a subject that I
find challenging and tricky, because of my days when I was
chairman of the board of NFA, when the first ATF draft
came out, it gets very tricky in trying to figure out where to draw the lines of
where the net gets cast as to what is an automated trading
system, and what isn’t. There are retail traders that
use a lot of the independent software vendor type systems
that employ pretty simple automated tools that quickly
fall into the category of automated trading system, depending on how the
wording of the rules is written. And so it gets very tricky and
very challenging to try to legislate
the We do a lot of low latency
automated trading and we spend copious
amounts of time going over our software
release protocols, our risk tools, as
Larry said there have been a lot of advancements, it still makes
me worry about it all the time. That some kernel of an operating
system goes down, and there’s something I agree with Larry that the
exchanges have made huge strides, and if
you ask the prop trading the principal trading firms, I think they are un must
that the best situation is better and
better the exchange tools get, not in any way to put the
responsibility on the exchanges, it’s the trading companies who
use the tools, it’s their responsibility to manage the
risk. But as those tools become more and more granular down to the tag 50
at the CME level, and the CMEs made great strides income able
to calculate what in the future business over in the equities business is buying power with
their GCC limits, or per product risk
limits, and that those tools continue to get better and
better, it’s very competitive, as we’ve talked about today, the
technologies and the speed of these things is breath taking. But it’s provided tremendous
value and liquidity to the U.S. treasury market to try to
finance all of that debt, the liquidity is deep. And so those tools have become
better and better, and the best setup I think for everybody is
if the exchanges continue to do that. That’s the ideal
situation. Because legislating, while
there’s a real temptation to do that, because it scares all of us, it gets very
tricky in exactly how to do that. So it’s a tough balance,
and it’s a challenge, and it’s tricky, and it’s a very good
public policy debate.But I do tend to think that these
technologies, as the exchanges have gotten better and better,
in the futures markets — I’m not sure about
all those venues that Larry was talking about in the equity
markets that make it so complex, but because I’ve got central
order books, the technologies have come a long way.>>Daniel Gorfine: Thank you.
Last comment Mr.>>Yeah I complete agree with
the comments from Larry and Mr. Hehmeyer over here, I’d like to
emphasize the standards rrnt to build perfect systems, hardware
and software will fail and always fail, I think the
standard should be to ensure that market participants have a
framework around controlling and mitigating those problems so
that they don’t, you know, have external impacts.>>Okay thank you very much. So
I think based on the panel and the discussion, and the range of
potentially worthy items for this group to
explore, I would move that the TAC recommend to the commission
that it consider creating a subcommittee on automated
trading. Is there a second? Any questions or comments? All right, I will now call for
the vote on the motion. All those in favor of recommending
to the commission, that the commission consider creating a
subcommittee on automated trading, please say aye. All those opposed, please say
nay. Are there any abstentions? Okay, the motion carries. I’m
getting better at this procedure, here. So before we take a break, I
believe the chairman has a few remarks he’d make >>Just briefly, I’m not going
to be able to stay for the final session so I just wanted to give
a closing remark now if I could. And actually I wanted to give
sort of a correction, and observation. And a reflection.
And the correction is that the professor is from Vanderbilt,
not Yale, Larry Tabb. We’ve got enough grade there, I just
wanted to make a point.>>Larry Tabb: I’m sorry about that.
Crews Chris there was an observation I >>Chairman Giancarlo: There
was a — regulate ertion were heard as vampires in the house,
the I’ve heard regulators called a few things and I probably
called them some myself, but vampire is >>Point of order we don’t want
regulators to be like vampires, you let them into the house and
you can’t stop them. No, no I’m not accusing
regulators of There was one tym Estes referred
to he said technology has not yet turned back and I think he
was referring to bad behavior in the marketplace. I think he
said yet. I thought the yet was quite
remarkable. I must say, there’s a lot about
technology I think is absolutely remarkable. And transformational, and can
bring a lot of good. But I’m not sure ending bad
behavior sug can do. I think as long as there are human beings anywhere in the market,
our role as regulators is to try to find them, punish th them,
take them out of the marketplace. But the strange
sthing is they — they or their friends come back at some other
time. So I think the ongoing battle that regulators, and self
regulatory organizations and good players in the market, we
all have together, is to find those bad actors and take them
out. Unfortunately, they will always be there. If technology
limits the number of them, that’s great. But I think
that’s probably not one that at least I personally have on my
hope list. I think that the process of getting rid of bad actors still falls on
us as humans to find them and take them out of And I think as
long as there are markets they will attract — there’s an old story about the bank robber,
when asked why he robs banks he says well that’s where the money
is. Why do bad architects go in
markets, well maybe that’s where the money is and I think our job
as good actors is to I want to thank you all this has been a
great, great day today it’s been a great advisory committee
meeting this year by the new commission so I tip my
hat to to comishl Quintenz who put a lot
of time into and it falls on the heels of a great MRAC meeting
just the week before. Thank you all very much have a good>>
Thank you very much Mr. Chairman. With that we’ll take
a 10 minute break I want to try to stick to the schedule, it is
valentines Dade if folks want to head out after the last panel,
let’s return back at 3:25, if we can,
for our (Break. Okay, I would like to call the
TAC meeting back to order, and begin our final panel on
emerging trends and best practices with respect to
cybersecurity. Our panelists are Naeem Musa of
the CFTC and fill 30 s Schneck of
Promontory IBM.>>Good afternoon, thank you for
the opportunity to address the technology advisory committee as
Commissioner Behnam mentioned in his opening reactors, cybersecurity is an
all hands on deck exercise. We are laser focused on
protecting our data from cyber attacks, cyber
attacks continue to rise and getting
more sophisticated, it lies in the
combination of hacking, malware and social engineering. In my role as the chief
information security officer for the
commission, I meet monthly with our chairman to review or recent cyberincidents and
agency responses as well as review all recent cyberincidents and
agency responses. In addition to that, I meet on a
regulatory because is with the commissioners to discuss our
progress in our cybersecurity program and review with them new
trends in cyber threats. We have heard multiple panel
participants comment on cybersecurity risk of data transmission,
storage, hackable trading trrchs and transparency in virtual let’s, while we
believe technology is driven, innovation is and will continue
to enhance markets, we also recognize the paramount
importance of cybersecurity and data security practices. That can hepatosafeguard the
system underpinning such innovation. According to the Verizon 2017
breach, data breach report, private sector financial service
organizations were the most targeted victims. Representing nearly a quarter of
all confirmed breaches. While other sectors reported
more incidents, over 40 percent of incidents impacting financial
services organizations resulted in
breaches more than any other sector. Cybersecurity remains one of the
top compliance risks for financial firms. According to the same report, 88 percent of industry compliance
officers surveyed viewed cybersecurity compliance and identity theft as the most
challenging compliance topic. To echo the words of our
chairman, cybersecurity sin doubtedly the most important
single issue facing our markets today in terms of market
integrity and financial stability. We are all in this together, and
it’s critical that we work together
in a private-public partnership to
combat cyberthreats. CFTC seeks to leverage every resource available to us across federal
government and private industry. CFTC is constantly reviewing and updating our cybersecurity
protections to guard against the growing threat of breach, of
breaches. We can’t get comfortable or rest on our
laurels with our past successes of preventing or avoiding
breaches. CFTC takes nothing for granted. The cyberthreat is persistent
and ever — changing. It has rightly been said and
it’s not a question of if, it’s a matter of when, when a breach
will take police station. In addition to collaborating
with other financial regulating agencies, CFTC works closely
with the Department of Homeland Security and the institute of
technologies to bring in leading technologies and frameworks for
combating cyberthreats. We are committed to seek opportunities
to utilize artificial intelligence, machine learning,
and automation, to strengthen our ability to combat
the constant morphs nature of cyber
risk. The CFTC has adopted the NIST cyberframework as a model for
how we deliver cybersecurity services and how we ensure
protection of our assets. When adoption began in late 2016
our leadership team recognized the ability of the cybersecurity
framework to provide standards for managing and reducing
cybersecurity risks, organizing capabilities around five function tall areas of
identifying risk, protecting against it, just as
in the case of public and private organizations CFTC nneeded comprehensive
structure for making informed decision about risk, as well as
budget and strategic planning to assure we are prepared for the
road ahead. In addition, my team has
established a security metrics program to better inform stakeholders how across
our organization to promote awareness and structure for
making informed risk based services and managing
cybersecurity risks. These metrics provide additional visibility that go above and
beyond the federal reporting guidelines under fist ma. As a result, FISMA. As a result our team has
achieved a green scorecard under all of
cybersecurity framework. CFTC responsible for delivering
cybersecurity services within the office of technology is
functionally aligned with the cybersecurity framework. We are constantly assessing our
current state of maturity and the pathway for the desired
state of maturity for each service. Including planned
projects, acquisitions technology,
addition of personnel, and corresponding budget required to
achieve each roadmap. In September 2016, the CFTC un
muls adopted safeguards and cyberresilience standards for
clearinghouses, contract markets, swap execution
facilities and swap data repository. At a minimum all
organizations should focus on governance, risk
assessment, access rights and controls, data
loss prevention, vendor management,
training, As our CFTC chief Mikeal gill
has stated we recognize the great expertise in cybersecurity
within the firms that we And proactively seek ways to
affirm their front line defense on cybersecurity and sib
resilience. In addition our chairman has recently agreed to
chair a new board-level task force on cybersecurity for the
international organization of securities commission IOSCO. It
is critical that sib ur security be a priority for all
regulators, and we believe IOSCO can play an important role in helping market regulators
around the world improve their
cyberresilience practices.In conclusion, I am interested in
hearing from the committee on how we can better share
information and improve how we better collaborate around
this indiscriminate challenge that we all face. When I hear the words all hands
on deck, with regard to cybersecurity threats, I
interpret that as the need for all of us to proactively
collaborate and work together. Thank you.>>Daniel Gorfine: Thank you
Naeem. Phyllis?>>Phyllis Schneck: Good afternoon, I’m Phyllis Schneck I’m noting my
tame tag, I thank you for adding my middle name here, for life if
anyone uses my middle name they’re mad at me. I appreciate
being here I appreciate the opportunity to be justice
committee. I’ve been a nerd all my life
studied hypercomputing early on for tornado modeling how you use compute
powering, went into cryptography, I showed my father
how to hack a few machines. He said please use that gift well don’t do bad things, he gave me
a talk about ethics and 20 years later
here we are. I can tell you my first world in
technology, everything is connected including government to
government your electric grid is talking to your refrigerator, to
your car. Running into my most recent role running the
operational defensive cybermission for the U.S. government tund undercertificate
of Homeland Security, you can see the day-to-day submitted
incident response that any moment it’s not just if and
when, it’s happening. And I’ll keep this part brief because I’m
sitting between you and valentines day, but the idea is
as computers are connected, computers are not smart, they’re
just fast. They fetch instructions off what
we all call memory. And they just execute it, they
don’t stop and think should I do this, somebody else is supposed
to have thought that through. And we haven’t, we haven’t built
security into that. So now we face, it’s not about
cybersecurity, I tell people in 10 years I don’t want to have a
field of cybersecurity, it’s about everything else. It’s
about the financial world that you drive every day, the
electric wold that our power companies drive, and daily life.
And we’re using this great new technology and we should enjoy
it, but we need to protect it by design.And as you have
everything being a computer, you all drove in one here today,
everything is executing instructions without thinking
about what it’s doing next. So at some point, whether it’s how
those instructions get to the machine, how they’re put into
it, how they’re crafted or at the time of execution, somebody
needs to put some logic into that and/or — and I give you
and — look as they are being executed what does this mean.
You won’t stop everything, you won’t see everything, but you
should start to detect things, and that’s where we get into two areas, our current
threat landscape and what we can do about that. In my current role leading the
global security practice for Promontory
financial group and IBM that’s a longer mouth full than my full
name we look at how do you not only take the role of explines
but make that security. They’re not the same. You can check a
bunch of boxes and they’re very important boxes, to give that
you, they’re very important. But the bad guy knows what those
boxes are. We have to go above and beyond that to get to resill
ypbs. So we look at as an industry k
you are investing in regulating or on the other side in being
compliant, look at that from an area of how does this
make me more resilient. Anyone that tells you you’re going to
prevent a cyber attack or they can prevent one for you, run
away. It should be about how am I going to detect it and how am
I going to be resilient. I believe very much in
communicating with regulators early and often, I come from a
world in the nerd community, where in the cyberoperation
center we all speak Klingon I actually
know a few words as events develop we’re looking as an
industry when do those get reported up to scaou. it used to
be you’re trained not to go upstairs with that so you have
your full story. Now it should be, now we have new technology
to put data together faster, you have a better idea of when
something is happening and I urge our customers and everyone
I talk to, bring t up there fast templt learn to speak English,
and translate what you’re seeing in cyberops into English that
someone can say this might be an event, that not only I’m seeing
but all my colleagues are seeing and this
work in public private partnership, the financial
sector is phenomenal at sharing information that protects the
entire sec to. When you look at what people like to call the
internet of thition, everything is connected that thermostat in
your house, Alex a they’re listening to you. Not in a good
or bad way that’s up to you, but it’s all about what is the data
that’s being transported, do you have any idea of how it’s being
transported, and we bring these devices into our workplaces. Do
we understand what data could get on them. Do we understand
what they’re connected to, is there a router
in our closet that helps somebody operate that nobody
knows about that’s cuct connected to something that runs
something operational in the building. These are all issues
we’ve looked at, in my prior life we looked at control
systems inside federal buildings and we said what are the deangs
in a random router for example being somewhere that So not to put a Hollywood movie
on this, but understand that in every sector, financial
especially, everything is connected. You look at high
performance computing the world is going toward fast compute
power. So Dan, I promised Dan I would
mention a couple of the emerging technologies, such as your
favorite one I’ll get there in a moment but when you heard
algorithms mentioned in the previous manl panel I would push
back and say yes the algorithms are getting better and better
but sometimes you can’t undo them sometimes they’re
mathematical transforms. I have sat in that seat somebody
said how did you get to that number and you can’t. And the
industry has put together billions of data points of
threats that ewith see all over the world, threats that come off
the box that sits at the edge of your network that blocks the bad
guy from getting in, that box is able to tell your ops center and
the ops center of the company that built the box, and those
companies use the speed of computing to put that together
just like a weather forecast in my old
life in forecasting tornadoes what this is coming together to
look like. The only difference between cyberand weather is the
chaos. So the bad guy controls the
cyber, and everything else controls the weather. But the
cyberis actually harder because someone else can dictate it, So as all these things come
together, it’s going to be about the power of computing, the
power is that power of high speed trading making sure it’s
accurate not only with the length of wire and kind of fiber
but who is able to adjust it at a pace faster than you can
detect them, and last part is most important was the
resilience how does the sector bounce back from that and self
correct. On the algorithm side some of that maths can get torn
apart some cannot, you sparked my interest today that’s a huge
area that I think the community can look for you in a better
way. What point are reliable for the decisions that are being
made, at what point are we not. Some of it can be taken apart,
some of it can’t. Compliance is turning to
rivalians, it’s not only did you comeck the boxes I’ll say a few nice things about
NIST network because its birthday I believe was yesterday
believe it or not and that was built with the private sector
and government together. A little bit of humor the
private sector send the folks the
framework not that would sent to days and days of government meetings, they sent
really good smart people to build that because the way we
could get public and private to work together, forgive me
without regulating because that was problematic in many, many
sectors for a lot 0 reasons. Putting together the best of
technology with the best of what government needed and it
provides a very good, it’s not everything, but a very good
baseline that people tend to truly appreciate. Because it’s
understandable. The technology that Dan begged
me to raise is our friend quantum computing the big Q. People will say that it will
outdo blockchain, there are a few claims there is a little bit
of truth to the last one for those who want a little bit
of nerd dom at the end of the day, the idea is you have an be
in multiple states at once. They actually qual call it a Q
bit in the community they don’t know what state it’s in ever.
To run through every possibility instead of having to find one in
several billion, at that point there’s at some point in the
state of that Atom that it will click, and you’ll find the
possibility, and that will literally enable you to, quote,
guess a key, I’m being very simple here, or and if you can
find a mathematical number, then you break cryptography. Now, we are quite aways from
that but not decades I would say about one decade some of this is
already being implemented that doesn’t mean you don’t encrypt
things from now until 10 years from now and say I give up, it
says please do incrept things when the bad guy comes in and
takes cases and cases of your data home to another planet you’ll be the one on the news in
a says great, but they got nothing because it was incrept inincripted we need
to understand these new foltion. Blockchain, cript crens
currencies, ability to spin up computers all over the world and
use their power to create numbers that equate to
currencies because it’s valuable it’s not at question of which
currency it’s a question of how the technology is applied. My personal opinion the best use
of the ledgers which is what they call blockchain is in
tracking so anyone who is slow and uses one of the airlines that uses a track my bags, I’m
not as angry when I land because I knew the bag didn’t get on the
plane but at he’s gits it’s getting tracked.They’re doing
the same things with vegetables now. Finding the E. coli in the
vegetables they do it in hours instead of weeks. They do it in shipping,
expanding a lot of the work all based on computing power and ability to track the
cryptocurrency, because it’s compute ag lot of computing
power to do fancy math for you to claim not only the
transaction but the actual currency. All of that is pretty much
already here so my message is threefold one is let’s continue
working with you to take compliance to maintaining
resilience, use the investment that customers and financial
institutions are making in compliance to actually be
cyberresilient and I would demand that of them as a rgher and many of them already
are and it’s nice to see that because you’re helping. I think
the second part is look at the technology you’re using. Almost everything you buy today
is which could connected that’s my favorite phrase. Not
technical. But if you look at it it’s got a USB somewhere, if
you plug your cell phone in an airport where there’s a USB
stands please don’t ever do that again those are actually little
computers and they can transfer data right on to your foneds
just use the plug. Think of every device you have where it’s
connected. Last part, think about where data flies all over. The ability to compute so
quickly takes disparate datasets, and puts them all together now, and creates
newly available information, from bits of data that were
never available to be mined and analyzed before. And creates
all new issues and privacy and compliance, all of a sudden new
data and new information, new tracking, is available that we
never imagined. So security and the thought
leadership that you’re put nothing now needs to go into
everything we do going forward so we are ahead of the game.
And we’re close now but we need to be ahead of the claim with
quantum and blockchain and there will be many, many Thank you again very much for
having me here and the work that you do.>>Daniel Gorfine: Thank
you so much and thanks to both of our panelists. I’d like to open it up again to
questions and discussion items. Hopefully that’s not me, that’s
making that noise. Phyllis — we’ll see if that
helps it. It does. Okay, and I’ll just kind of tee
up for consideration and picking up on this last thread, with all
of these emerging areas and he merging technologies that we’ve
talked about today, are there unique cybersecurity concerns
and approaches that you all are thinking about. And on a
related point, you know, are there emerging best practices
that are either available, that could be kind of consolidated by
a group like this, or could a group like this hepatocreate
best practices in some of these nascent areas. So I’ll throw
that out there, and any other thoughts or questions, as well.>>I obviously talk too much. Thank you guys very much, those
were great presentations, this is an area that’s near and dear
to my house and actually this is one of those cross-cutting
concerns that basically brings together everything we’ve talked
about so far today. One of the things I’m doing is
building a whole machine learning as a service platform,
and a lot of the data that we want to supply to that platform
is all this data that’s coming in from all these participants,
you know, thousands of connections from participants
all over the industry.And we need to really protect that
data, but we also want to design predictive analytics and all
kinds of other things. So how do you defend against leakage of
that data or people seeing things So we put in an awful lot of
thought into how we redact parts of the data, how we anonymize data, how we
prevent people from anonymizing the data on two different days
and then being able to correlate things across anonymized
datasets examine to do traffic analysis and say this one is
obviously that bank, or — so it’s an enormous cross-cutting One of the things that always
jumps out is, you know, any security measure that you take
is great until you learn that it doesn’t work anymore, that
somebody has figured out how to get past
that. So no one security measure
works, you have to have multiple layers and
multiple ways to audit. And as you said it’s not a matter of
necessarily stopping it always, but detecting that it’s happened
and being I guess that’s what I wanted to
mention.>>Daniel Gorfine: Okay, any
other questions or comments from the members? Oh I’m sorry, yes
Mr.>>Just wanted to kind of reiterate I appreciate the nerd
community is on our side here, and trying to help out. But
again as we talk about kind of the move to a distributive
ledger technology and a lot of the work that lab
CFTC is doing is to bring some technology partners into the
fray as well I think there’s a whole lot of data that’s going
to be available to people so I think we need to be mindful of
the entitlements to that data, how we access it do we anonymize
it, the enormous benefits across operational settlement and
regulatory reporting are immense, so I think we have to
go there but we just need to be extremely mindful from the start
about how we set these standards, and that’s got to be
a top consideration, not something we think about after
the fact. So I think definitely a
subcommittee that is focused on this topic would be definitely
recommended. Well, I will pick up on your
recommendation there and based on this discussion I would move
that the TAC recommend to the commission that it consider
creating a subcommittee on cybersecurity. Is there a
second? Are there any questions or
comments? Okay, I now call for the vote on
the motion. All those in favor of recommending to the
commission that the commission consider creating a subcommittee
on cybersecurity, please say aye. All those opposed, please say
nay. Are there any abstentions? Okay, with that, the motion
carries. So I’d like to thank very much
our panelists, and at this point I’d like to turn this back to
Commissioner Quintenz.>>Thanks very much Dan, first congratulations on your
proficiency with robots rules of order did a great job tidy guess
I’d turn it over to Commissioner Behnam for any closing remarks.>>Rostin Behnam: Thank you
first and foremost thank you Commissioner Quintenz for your
leadership holding this meeting, echoing the chairman’s comments
a really fantastic day. Where I think we all at least speaking
for myself learned a lot and raised a lot of new questions
which we have to address in the months and years ahead. Thank you to Dan Gorfine, from
my experience with MRAC and working
with Alicia Lewis, the DFO there, we
know this is a lot of work, so congratulations and well done.A
couple things I just want to point out and I’ll keep this
brief, a couple analogies that were raised first of all I love
the baseball analogy but I think it’s worth pointing out one of
the gentlemen who was used en actually got sort of at least
suspected or was alleged to have used performance enhancing
substances, so a fresh reminder that as we create these
incubators as we create these labs, as we create these
sandboxes it’s important that we keep — you know, bad
actors, manipulation, customer protections, fraud, all on mind.
Which I know we do, it’s a priority for everyone at this table, but
we have to beat that drum over and over again so that we
support innovation, but also do it in a responsible way so that
above all else, integrity is maintained and The other analogy which I think
Charley mentioned was that the journey analogy, which I think
is great in the sense that we have to do this together.
Obviously, public-private partnerships were mentioned
several times. One thing that I thought of, though, is as we do
this together, which we should, and I think the chairman has
demonstrated his willingness to do this time and again for the
past few months, I think it’s important at least from my standpoint that you all coalesce
around some core principles and some core ideas. Because there
clearly will be differences in terms of your business models
and what your interests are, and I think it was mentioned earlier
that none of us want to be picking winners and losers here. So for us to be good
policymakers and I think for the Hill as well, because eventually
I think we all believe they’re going to have to get involved in
terms of statutory changes, it’s going to be important to
understand that there are some baseline principles that you all
agree on, that may not be the optimal, but as a standard
there’s a best for the industry moving forward. So I think that’s an important
element to think about, as a lot of ideas were thrown around
today, productive ones, good ones, but things that I think we
have to digest and sort of analyze before we make final
decisions on how to move forward.So with that, thanks
again to everyone on the panel, look forward to
2018 and more important discussions. A lot of great
subcommittees brought into order today so looking
forward to discussions and the recommendations that come out of
there. So thank you again, Commissioner Quintenz, and happy
valentines day.>>Thank you Commissioner Behnam. And I’d like to chairman Giancarlo for
spending such a chunk his day with us and
Commissioner Behnam for sharg wus I’d like to
say technology is a bipartisan issue but the work of this
committee and I think the work of all the advisory committees
isn’t own by us, it’s owned by you.And we look forward to all
of the good work, and it’s going to be a lot of work, but all of
the good work that’s going to be done over the next one, two, six, 12 months, possibly even
longer, arnt some great conversation that we had today. I’d like to bringum something
that I think tym mentioned about hope versus reality. And I think it’s fine to talk
about hope and the possibility, and it gets us excited, but at the end of the
day it comes down to work, and understand whag we need to do to realize, you
know, that vision. And I’m excited that I think at
least in some ways this is where that work is going to get done.
So thank you all for your participation, and for your
future commitment to getting into that work. So and Dan,
great job. Thank you. So I’ll turn it back over to you.>>Brian Quintenz: For those of
you who were too shy to eat cup capes when you were on camera
feel free to do that when >> Thank you all again for
attending this TAC meeting it’s been incredibly interesting and
productive, with that I’d like to wish everybody a happy
valentine’s day and this meeting is adjourned.

Leave a Reply

Your email address will not be published. Required fields are marked *