November 24, 2015
The electronification of trading over the past 30 to 40 years and the advent of exponential digital technologies have transformed financial businesses, markets and entire economies, with dramatic implications for capital formation and risk transfer. In U.S. futures markets, we see this change most presently in the area of algorithmic or automated trading that now constitutes up to seventy percent of regulated futures markets. Automated trading can lower transaction costs while increasing trader productivity through greater transaction speed, precision and sophistication. For many markets, automated trading brings trading liquidity, broader market access, enhanced transparency and greater competition.
At the same time, automated trading presents a host of potential new challenges. They include increased risk of sudden spikes in market volatility and “phantom” liquidity arising from the sheer speed of execution, potentially flawed algorithms and position crowding. They also include the risk of data misinterpretation by computerized analysis and mathematical models that increasingly replace human thought and deliberation. Legal scholars raise important questions about the viability of traditional market regulation in automated trading markets.
How markets and market regulators adjust to this change from human to automated trading will be extremely important. It requires delicate balancing. To ensure vibrant, accessible and durable markets, we must cultivate and embrace new technologies without harming innovation. Without a doubt, there must be effective safeguards of market integrity and credibility, but those safeguards should not bar promising innovation and continuous market development.
In turning to Regulation Automated Trading (“Regulation AT”), I acknowledge that my staff and I had dozens of issues and concerns that we brought to the attention of the Division of Market Oversight. While they were responsive to a few small topics, many other issues require much further attention and consideration that I will summarize in this statement.
Still, after reading through the almost five hundred pages of this proposal, I am left with one major question that I still cannot answer.
That question is: does this proposal sufficiently benefit the safety and soundness of America’s futures markets so as to outweigh its additional costs and burdens?
I wish the answer was clearer.
I have three main concerns with Regulation AT. First, some of the requirements of the proposal appear to be window dressing. That is especially the case in its requirement for development and implementation of risk controls and related testing standards that the industry has already widely adopted.
Second, I am concerned about the high costs and burdens of this proposal, especially on small market participants. I am especially concerned about its requirement that registrants hold their proprietary source code in data repositories available for inspection by the Commission or the U.S. Department of Justice at any time for any reason.
Third, I question the regulatory inconsistencies regarding the market participants that must comply with this rulemaking.
For these reasons and others, I have serious doubts about today’s proposed rulemaking.
Last November, I delivered a speech at the U.S. Chamber of Commerce where I set forth six principles that I would follow as I evaluate financial market regulations.1 As part of those principles, I proposed the “SMART REG” standard to help analyze whether CFTC rules actually solve for real problems and promote the U.S. economy and the American markets.2 I struggle to see how Regulation AT passes the SMART REG standard.
Nevertheless, I want to hear the views of market participants on this proposal. I will evaluate any final rule based on the SMART REG standard and thereafter determine whether to support or reject it.
I will explain my areas of concern.
I. Necessity of Regulation AT
It is hard to identify exactly what issue in automated trading Regulation AT is designed to address. The agency is basically playing catch-up to an industry that has already developed and implemented risk controls and related testing standards for automated trading. Regulation AT describes the extensive best practices and recommendations for automated trading issued by industry organizations and notes that the majority of industry participants are following such best practices. Regulation AT simply codifies industry best practices in many respects, but does not go as far as current industry efforts. As such, the Commission admits that many of the benefits of this proposal are already being realized in the marketplace. In reality, current industry standards on automated trading have well surpassed Regulation AT in many areas.
It is clear that the industry has long been at the forefront of creating market solutions for risk controls in automated trading well before any regulatory mandate. As I recently stated, I favor this type of ongoing bottom-up market-driven approach to risk controls for automated trading.3 Given the industry’s leadership role and the fact that Regulation AT simply codifies a small subset of industry best practices, while adding heavy compliance burdens, I question the necessity and value-add of this proposal.
The staff partly justifies the proposal as necessary to ensure market integrity given the risks of automated trading. As support, Regulation AT illustrates examples of recent disruptive events in automated trading. However, the dearth of incidents in the futures market seems to indicate that current industry solutions are working well. Regulation AT only cites three U.S. disruptive automated trading events in the past five and a half years and two of those events occurred in the equities market, obviously outside of our jurisdiction. In addition, the equities market events occurred despite the Securities and Exchange Commission (“SEC”) having implemented some reforms related to automated trading.4 Thus, I question whether Regulation AT will in fact reduce future disruptive events and enhance market integrity.
As further support for market integrity, the preamble asserts that the proposal may limit a “race to the bottom,” in which certain entities sacrifice effective risk controls in order to minimize costs or increase the speed of trading. In this, the proposal betrays a naïve misunderstanding of elementary micro-economics. Market participants have every economic incentive to implement effective risk controls to prevent the loss of their capital and being forced out of business. That is why the industry has been a leader in best practices for automated trading, including development of risk controls and related testing standards. This ongoing bottom-up market-driven approach to risk controls for automated trading has raised, not lowered, the standards.
Several commenters cited in Regulation AT supported a principles-based approach to regulation citing the need for flexibility because each market and the participants in those markets are different.5 These commenters also noted that the Commission already has robust regulations in place to address the risks of automated trading.6 Tweaking the Commission’s existing regulations7 in line with a principles-based approach may be a better way to build upon ongoing industry efforts regarding automated trading, while reducing the compliance burdens of Regulation AT.
I invite comment on the necessity of Regulation AT and on other approaches to automated trading that support – rather than burden – ongoing industry efforts.
II. Costs of Regulation AT versus the Benefits
I am concerned about the costs of Regulation AT, especially on small market participants. The Commission tries to downplay the costs of this proposal because in many respects it simply codifies industry best practices and many market participants are already following such practices.8 The proposal also repeatedly asserts that the rules are flexible seemingly in an effort to highlight its low burdens. However, in reality, Regulation AT adds compliance, reporting and registration requirements, and establishes designated contract market (“DCM”) and registered futures association (“RFA”) review programs. These additional requirements will certainly increase costs to all market participants engaged in Algorithmic Trading that are subject to this proposal.
A. Small Market Participants
The costs of this proposal may disproportionately impact small market participants. While Regulation AT raises this concern and asks questions in this regard, at the same time, the proposal dismisses the possibility that it will capture many small entities. I am not so sure that will be the case as the definition of Algorithmic Trading is very broad and would appear to capture market participants using off-the-shelf type automated systems or simple excel spreadsheets to automate trading.9 If that is the case, then this proposal could capture, for example, a small proprietary trading firm, a small commodity trading advisor (“CTA”) or a rural grain elevator company that uses simple automation.
Regulation AT would add numerous costs to small market participants and raise barriers to entry. Small market participants may be less likely to employ risk controls consistent with Regulation AT so they would incur costs to develop or purchase such risk controls. They would also incur costs to hire additional employees to develop and implement policies and procedures for the development, testing, monitoring and compliance of their Algorithmic Trading systems. Small market participants would have to hire additional employees to continuously monitor their Algorithmic Trading systems on a real-time basis. They would incur costs to annually prepare and submit a pre-trade risk control compliance report to each DCM on which they trade. Furthermore, the proposal would add costs to small market participants given the required registration with the Commission and an RFA. That sounds like a whole lot of extra costs to me for a “principles-based” non-prescriptive rule.
The proposed rule admits that the Commission does not know how many small entities this proposal will affect – unfortunately, a common theme for CFTC rules when discussing costs and burdens on the marketplace. I am disappointed that the Commission did not get a better sense of the potential universe of small market participants that may be impacted by Regulation AT. To this point, I am very interested to hear estimates of costs Regulation AT will impose on smaller market participants and how it will impact their ability to conduct business.
Interestingly, the proposed rule also asserts that a technological malfunction or error in a very small proprietary trading firm’s algorithm could have a significant, detrimental impact on the market despite providing no evidence to support this claim.10 I invite commenters to weigh in on this issue. I am also interested to hear comments on whether the proposed rules make sense for those market participants using off-the-shelf type automated systems or simple excel spreadsheets to automate trading, especially the rules around development, testing, monitoring and compliance of Algorithmic Trading systems.
B. Overlapping Requirements and Duplicative Costs
Regulation AT contains a potential overlap in some requirements and duplicative costs in that it requires AT Persons to register with an RFA and, at the same time, to be subject to reviews by DCMs. The National Futures Association (“NFA”), the only RFA at this time, will need to hire additional employees to establish and maintain a program for Algorithmic Trading systems. The preamble to Regulation AT also contemplates that NFA would conduct routine examinations of its members to ensure that they are complying with NFA rules. This requirement translates into additional costs that will be passed down to NFA’s members. Regulation AT notes that NFA is the frontline regulator and is well-positioned to address rules and issues related to Algorithmic Trading as market conditions and technology develops.
However, it seems that DCMs have the most intimate knowledge of the markets and their participants trading in those markets. DCMs have been at the forefront of creating market solutions for risk controls in automated trading, along with testing and certification of automated systems.11 In this regard, Regulation AT requires AT Persons12 and their clearing member futures commission merchants (“FCMs”) to submit annual reports and policies and procedures regarding their Algorithmic Trading to all DCMs on which they trade. DCMs must establish a program to review these reports and procedures and provide feedback, including any deficiencies in participants’ pre-trade risk control settings or calibrations.13 AT Persons and their clearing member FCMs must also keep, and provide upon request to DCMs, books and records regarding compliance with the proposed rules. DCMs must review these books and records as necessary.
Although the preamble states that the NFA and DCM requirements are not intended to create conflicting obligations, I am afraid that the lack of clarity provides a potential to subject AT Persons to some duplication. As noted above, DCMs already have standards for risk controls, testing and certification of automated systems, but Regulation AT requires NFA to address these topics in its program. Regulation AT also discusses reviews for both NFA and DCMs. Duplicative requirements would add unnecessary costs that would be especially harmful to small market participants.
I am interested to hear from market participants if Regulation AT provides enough clarity on this issue or if the Commission should provide further detail. I am particularly interested to hear comments on the requirement for market participants to register with NFA and to be subject to NFA’s program for Algorithmic Trading systems. In light of DCMs’ existing efforts on risk controls and testing, is such a requirement necessary or are DCMs already serving as the frontline regulator? Would NFA serve a useful role in setting consistent standards across all markets or do DCMs need flexibility in setting rules because each market and the participants in those markets are different?14 I also invite comment on alternatives to the requirement that AT Persons and their clearing member FCMs prepare and submit annual reports to DCMs and DCM reviews of those reports. One possibility is to require AT Persons and their clearing member FCMs to conduct self-assessments (like FINRA requires) and only require submission to DCMs upon request.
C. Source Code Repository and Commission Regulation 1.31
Source code is the intellectual property of AT Persons representing their current and future trading strategies. Source code of AT firms is unlike traditional trading firm information in that it reveals not what positions are held in the past or present, but what positions the firm intends to buy or sell in the future upon specified market events.
I am particularly concerned that Regulation AT requires that each market participant keep a source code repository for algorithms and make it available for inspection to any representative of the Commission or the U.S. Department of Justice for any reason.15 Currently, the federal government may only obtain such sensitive information through a subpoena. Regulation AT dramatically lowers the bar for the federal government to obtain this information.
I am unaware of any other industry where the federal government has such easy access to a firm’s intellectual property and future business strategies. Other than possibly in the area of national defense and security, I question whether the federal government has similarly unfettered access to the future business strategy of any American industrial sector. Does the SEC require such access from its registrants? Do other agencies in the federal government have ready access to businesses’ intellectual property and business strategies?
I am unclear why either the Commission or the U.S. Department of Justice needs access to source code information without a subpoena, especially the Justice Department, whose only use for such information would be in criminal proceedings. Does today’s rule proposal presume that the use of automated trading technology makes a trading firm more likely to engage in criminal behavior than a manual trading operation?
There is strong reason for concern about maintaining the confidentiality of this source code. As we all know, the federal government has a poor track record of keeping sensitive information secure from cyberattacks and other data breaches. Any data breach of this information would be devastating for such entities and, potentially, for the safety and orderly operation of U.S. markets. Imagine the harm that could be caused to U.S. financial markets, if cyber terrorists or other belligerents were able to get their hands on this technology the same way some of the U.S.’ most important industrial, military and other sensitive data have been hacked. I question the need for this new requirement and request commenter feedback on this issue.
In addition to my concerns above, I previously expressed reservations about Commission regulation 1.31 in the proposed rulemaking on Records of Commodity Interest and Related Cash or Forward Transactions.16 Commenters to that proposed rulemaking stated that Commission regulation 1.31 is technologically outdated and compliance with the rule is overly burdensome, infeasible and costly.17 Managed Funds Association, the Investment Adviser Association and the Alternative Investment Management Association even petitioned the Commission to amend Rule 1.31 back on July 21, 2014.18 Unfortunately, the Commission has not acted on this request.
Regulation AT’s requirement that source code repositories must be kept and made available for inspection pursuant to Commission regulation 1.31 will impose unnecessary costs and burdens on AT Persons. Given the voluminous comments that the staff has received on the unworkability of Rule 1.31, I am surprised that Regulation AT would subject source codes to this rule. As an alternative, the Commission should consider allowing AT Persons to keep source code repositories in accordance with their own reasonable and secure internal recordkeeping procedures. I welcome comments on the costs of Commission regulation 1.31 in this regard.
Finally, I would like to note that currently unregistered market participants who will now be required to register under the revised floor trader definition may be subject to heighted record keeping requirements under proposed Commission regulation 1.35.19 Proposed Rule 1.35 states that a member of a DCM that is not registered or required to be registered with the Commission in any capacity would not have to keep (i) records of transactions in a manner that is searchable or allows for identification of a particular transaction20 and (ii) text messages related to those transactions.21 If the Commission finalizes Rule 1.35 as proposed, Regulation AT’s registration requirement would increase the burdens under that rule. I invite commenters to provide feedback on the intersection of Regulation AT and Rule 1.35.
D. Other Costs
I would also like to obtain industry input on the following costs of Regulation AT:
1. The costs on FCMs under proposed Rules 1.82 and 1.83, especially the requirement that an FCM prevent or mitigate an Algorithmic Trading Disruption for its AT Persons.22 I have previously expressed concerns about the harm caused to the FCM industry by the heightened cost of regulation, so I am especially interested to hear comments in this regard.23
2. The costs to DCMs to establish and maintain a program for the review and evaluation of compliance reports and books and records of each AT Person and their clearing member FCMs trading on the DCMs, as required under proposed Rule 40.22.
3. The ease and costs for DCMs to generate and publish self-trading statistics, as required under proposed Rule 40.23(d).
E. Costs versus Benefits
Based on all the costs described above, Regulation AT does not seem to be a non-prescriptive, low-burden rule that simply codifies industry best practices as the proposal asserts. It goes much further and, I fear, does greater harm. While Regulation AT does recognize industry best practices with respect to several risk controls, it adds prescriptive compliance, reporting and registration requirements and establishes overlapping and duplicative DCM and RFA review programs of questionable value. Given the industry’s extensive efforts to date, I question whether the costs of Regulation AT actually justify the benefits. The principles-based approach that I discussed above may be as effective and less costly than Regulation AT’s approach. I invite commenters to provide feedback regarding the costs and benefits of Regulation AT and the specific points I raised above.
III. Regulatory Inconsistency of Regulation AT
I would like to note three regulatory inconsistencies in Regulation AT. The staff proposes to amend the definition of floor trader24 in order to register currently unregistered persons using direct electronic access for algorithmic trading on DCMs.25 The preamble to Regulation AT states that in 1993, when the Commission finalized rules regarding the definition and registration of floor traders, the Commission decided to include as floor traders only those traders operating on the trading floor of an exchange. However, in that 1993 rulemaking, the Commission stated that certain traders using electronic trading systems come within the floor trader definition. Back then, the Commission took a technological approach to the definition of floor trader.
Today, Regulation AT is taking that same approach and is proposing to register persons using direct electronic access for algorithmic trading, but not those using manual means. I am not clear on the rationale for this technology driven distinction to registration (as the preamble does not articulate one) when the proposal acknowledges that manual trading also poses risks. Several commenters cited in Regulation AT also noted the importance of risk controls for manual and automated trading systems.26 I invite industry comments on this issue, notwithstanding my above concerns about the registration requirement.
Another regulatory inconsistency is that Regulation AT only captures floor traders who use direct electronic access for algorithmic trading, but it captures all existing registrants, such as FCMs, swap dealers and CTAs regardless of whether they use direct electronic access for algorithmic trading. Again, Regulation AT does not articulate a reason for this inconsistency and I question its logic. I invite comment on this issue, including whether, for existing registrants, the proposal should only capture those using direct electronic access.
Finally, Regulation AT only applies to trading on DCMs and not on SEFs. Regulation AT justifies this distinction by stating that compared to DCMs, SEFs and SEF markets are newer and less liquid and have less automated trading. However, DCMs can also list swaps and Regulation AT applies to that trading. In this regard, Regulation AT may disadvantage DCMs who list swaps as compared to SEFs. I welcome comments on this competitive disadvantage, including whether Regulation AT should exclude from its scope swaps listed on DCMs.
IV. Other Comments on Regulation AT
I also invite industry comment on the following issues:
1. Whether the Algorithmic Trading Compliance Issue definition in proposed Commission regulation 1.3(tttt) is necessary. If a major reason for Regulation AT is market integrity then it seems the Algorithmic Trading Disruption definition is sufficient. Furthermore, if an AT Person violates a rule or regulation it will be liable so the Algorithmic Trading Compliance Issue definition appears unnecessary.
2. Whether the definition of Direct Electronic Access in proposed Commission regulation 1.3(yyyy) should be harmonized with the definition in Rule 38.607.27
3. Whether several of the proposed rules that require periodic review of compliance measures or regular testing of Algorithmic Trading systems open up AT Persons to liability risk. For example, proposed Commission regulation 1.80(f)28 requires each AT Person to periodically review its compliance with the pre-trade risk control requirements to determine whether it has effectively implemented sufficient measures reasonably designed to prevent an Algorithmic Trading Event. What happens if market conditions change rapidly between periodic reviews and the AT Person’s risk controls are no longer sufficient to prevent an Algorithmic Trading Event? Is the AT Person now liable for a violation of Commission rules? Will this periodic review become a continuous review in order to avoid liability?
While I am pleased that Regulation AT provides flexibility in setting risk control parameters and does not require the pre-approval or pre-testing of algorithms, the proposal appears to add many burdensome compliance costs and does not adequately take into account small market participants or the work of the industry in developing algorithmic trading risk controls and related testing requirements. Rather than duplicating their efforts and adding additional burdens, the Commission should look to support and enhance ongoing industry progress. On the other hand, I am highly concerned about Regulation AT’s several significant inconsistencies and its extraordinary requirement that AT source codes be placed in government accessible repositories.
Overall, I have a great many concerns with Regulation AT. Most principally, I struggle to figure out if it will benefit the safety and soundness of America’s futures markets enough to outweigh its additional costs and burdens. Its purpose must not be to allow a Federal regulator to say that it has “done something” about computerized trading in response to media headlines, best-selling books or political campaign agendas. The development of automated trading is too complicated and too important to be addressed with such superficiality.
For my part, I will carefully review thoughtful comments from market participants and the public. I will measure my support for any final rule against the SMART REG standard.
1 Remarks of CFTC Commissioner J. Christopher Giancarlo before the U.S. Chamber of Commerce, Re-Balancing Reform: Principles for U.S. Financial Market Regulation In Service to the American Economy, Nov. 20, 2014, http://www.cftc.gov/PressRoom/SpeechesTestimony/opagiancarlos-2.
2 Id. The “SMART REG” standard follows whether new CFTC regulations S- Solve for real problems, not anecdotes of bad behavior; M- Measure success through a rigorous cost benefit analysis; A- Advance innovation and competition through flexible rules; R- Represent the best approach among alternative courses of action; T- Take into account evidence, rather than assumptions; R- Realistically set compliance deadlines; E- Encourage employment of American workers; and are G- Grounded in law.
3 Keynote Address of CFTC Commissioner J. Christopher Giancarlo before the 2015 ISDA Annual Asia Pacific Conference, Top-Down Financial Market Regulation: Disease Mislabeled as Cure, Oct. 26, 2015, http://www.cftc.gov/PressRoom/SpeechesTestimony/opagiancarlo-10.
4 See Regulation AT Preamble, Section II.C.1.: “Background on Regulatory Responses to Automated Trading.”
5 ICE Comment Letter at 1-3 (Jan. 17, 2014); Katten Muchin Rosenmann Comment Letter on behalf of Gelber Group at 5, 20 and 22-24 (Dec. 9, 2013).
7 See e.g., Commission regulations 1.11(e)(3)(ii), 1.73, 23.600(d)(9), 23.609, 38.255 and 38.607.
8 I also note that the Commission uses old compensation data from 2012 in calculating the costs of Regulation AT, which underreports these costs estimates.
9 See definition of Algorithmic Trading in proposed Commission regulation 1.3(ssss).
10 See Regulatory Flexibility Act section of Regulation AT.
11 E.g., CME Comment Letter at 25-26 (Dec. 11, 2013) (discussing CME’s two testing environments for its users and its certification requirement).
12 AT Person is defined in proposed Commission regulation 1.3(xxxx) and captures the persons subject to Regulation AT, including existing Commission registrants engaged in Algorithmic Trading and the newly expanded definition of floor trader.
13 Proposed Commission regulation 40.22(c).
14 See supra note 5.
15 Under Regulation AT, in accordance with Commission Regulation 1.31 (17 CFR 1.31), AT Persons would have to make their source code repository available for inspection to any representative of the CFTC, in addition to the U.S. Department of Justice.
16 Records of Commodity Interest and Related Cash or Forward Transactions, 79 FR 68140, 68148 (proposed Nov. 14, 2014).
17 Managed Funds Association Comment Letter at 4-7 (Jan. 13, 2015); Commodity Markets Council Comment Letter at 5 (Jan. 13, 2015); SIFMA AMG Comment Letter 5-6 (Jan. 13, 2015).
18 Managed Funds Association, the Investment Adviser Association and the Alternative Investment Management Association, Petition for Rulemaking to Amend CFTC Regulations 1.31, 4.7(b) and (c), 4.23 and 4.33 (Jul. 21, 2014).
19 Records of Commodity Interest and Related Cash or Forward Transactions, 79 FR 68140 (proposed Nov. 14, 2014).
20 Id. at 68146, Proposed Commission regulation 1.35(a)(3)(i).
21 79 FR at 68146, Proposed Commission regulation 1.35(a)(3)(ii).
22 Proposed Commission regulation 1.82(a)(i).
23 Statement of Commissioner J. Christopher Giancarlo for the Market Risk Advisory Committee Meeting, Jun. 1, 2015; http://www.cftc.gov/PressRoom/SpeechesTestimony/giancarlostatement060115.
24 17 CFR 1.3(x).
25 Another requirement is that the person must be trading for their own account.
26 E.g., CME Comment Letter at 43, 44 (Dec. 11, 2013).
27 17 CFR 38.607.
28 See also proposed Commission regulations 1.81(a)(1)(iii), (a)(1)(iv), (a)(2) and (c)(2)(i) for further examples.
Last Updated: November 24, 2015