Regulation Ahead: Advice and Options for Automated and High-Frequency Traders
By Zachary J. Ziliak, Mayer Brown LLP
Have no doubt: From one source or another, new rules are coming that will govern trading firms’ use of technology. The Securities and Exchange Commission took a major step in that direction on March 7, when it proposed Regulation SCI (“Systems Compliance and Integrity”).1 The “Systems” in that title refer to the “automated markets and computer-driven trading”2 that have received much journalistic attention in recent months, most of it negative. While that particular regulation is aimed primarily at securities exchanges and alternative trading systems, many of its rules could apply equally to other market participants, and the Commodity Futures Trading Commission and Congress have likewise dropped heavy hints of additional controls over automated trading to be added soon.
Proposed Regulation SCI points toward a more exacting and intrusive look into trading software than has been applied to date under the optional Automation Review Policy—most immediately for “SCI Entities” and potentially for broker-dealers later.3 Similarly, for investment advisers, commentary from hearings and the practice of the SEC’s new “quant audits” suggest that both models themselves and the compliance procedures that surround their development will be on the agenda. Moreover, while high-frequency trading (HFT) gets the most attention from media and regulators alike, many concerns highlighted in Proposed Regulation SCI and hearings before regulators conceivably apply as well to other automated trading companies, quant shops, and even traditional traders who simply rely on computerized execution algorithms.4
This article surveys the state of play for automated, algorithmic, high-frequency, and quantitative traders in the U.S. regulatory environment. Specifically, the sections below summarize (1) the concerns that motivate additional regulation in this area, (2) the actions taken by Congress and regulatory agencies over the past year in relation to high-frequency trading, (3) topics for likely future regulation in this area, and (4) actions industry participants are taking in response.
HFT: Benefits and Concerns
Contrary to the image commonly presented in the media, HFT is neither revolutionary nor necessarily harmful. Investors have sought for centuries to acquire information before their competitors and thus to gain a trading advantage; technology has simply advanced to a stage where those advantages are measured in microseconds rather than hours. Today, across both high- and low-frequency strategies, “virtually all equity trading in the US is automated,”5 and the same can be said of many other commonly traded products. Numerous studies credit HFT with narrowing spreads, reducing execution time, and boosting liquidity,6 as well as potentially lowering volatility.7
At the same time, opponents of HFT can point to plenty of studies of their own that suggest that high-frequency trading does more harm than good.8 HFT is variously accused of costing investors a few cents on every trade, free-riding on the contributions of others to fund its excessive cancellations, leaving traditional market-makers with a disproportionate share of “toxic flow,” pulling out of the market exactly when liquidity is needed, clogging communication channels with orders that are not intended for execution, and hurting market quality in any number of other, similar fashions.
High-profile market disruptions have also led investors and regulators to question the reliability of electronic trading, especially HFT. On May 6, 2010, U.S. equity markets tumbled at an unprecedented rate, wiping out nearly $1 trillion in notional value in minutes before stocks regained nearly all their lost value that same day. The SEC and CFTC eventually traced these market gyrations to an automated trading algorithm that was instructed to sell 75,000 E-Mini S&P 500 futures contracts (worth about $4.1 billion) at a rate that depended on current trading volume in those E-Minis—only to have the algorithm’s own trading trigger follow-on HFT trades that drove E-Mini volumes higher, accelerating the algorithm’s sales so that all 75,000 contracts were sold in a matter of minutes.9 The programs behind both the underlying mass E-Mini sale and the high-frequency purchases and subsequent sales of those contracts were viewed as having contributed to the disruption. While that “Flash Crash” of 2010 remains the prime example of how interacting trading algorithms can suddenly trigger large market moves, Nanex, LLC has since compiled numerous smaller instances of what appear to be algorithm-driven cascades.10
Such events point out how trading algorithms, even when working correctly, can interact with one another so as to create extreme events. Equally, the past year has delivered instances of trading algorithms and other programs not working correctly. On August 1, 2012, Knight Capital lost approximately $440 million in just 45 minutes as a result of problems with new trading software it rolled out.11 Further afield, computer problems that affected the Facebook12 and BATS13 initial public offerings do not appear to have stemmed from trading algorithms, but have been cited by some as contributing to a general decline in investor confidence in the integrity of modern electronic markets. For example, Senator Jack Reed, then the Chairman of the Senate Banking Committee’s Subcommittee on Securities, Insurance, and Investment, commented that “[t]aken together, these failures in electronic trading appear to be affecting investor confidence in the U.S. market structure.”14
In particular, popular opinion has singled HFT out for attack. Recent months have seen numerous articles warning of the risks of HFT or calling for its curtailment.15 A general lack of understanding of how HFT works and a common perception that favored companies enjoy an unfair and insuperable timing advantage over other investors16 contribute to the industry’s negative reputation. And when popular concern over market integrity is that focused, governmental action tends to follow.
Recent Governmental Response
While the U.S. arguably leads the world in HFT, other countries have been more aggressive in regulating such trading. Australia has issued rules that will require dark pools (trading venues that do not display resting orders to other market participants) to offer “meaningful price improvement” before internalizing trades and will force algorithmic trading companies to install “kill switches,” which can prevent rogue algorithms from sending any further trades to the market.17 A committee in the European Parliament voted 45-0 in favor of establishing a minimum resting period for orders of 500 milliseconds.18 The Australian Securities and Investments Commission recommended this same minimum resting period in March 2013, albeit only for small orders.19Canada imposes fees designed to discourage “quote stuffing,” a practice in which trading companies intentionally issue and cancel great numbers of orders in order to eat up market bandwidth and slow down other traders.20 Germany is looking to require HFT companies to register so that regulators can track the registrants’ trading activity.21
But while a few other countries may have been more active thus far in regulating HFT, the U.S. is not far behind. In particular, the Senate’s Committee on Banking, Housing, and Urban Affairs, the SEC, and the CFTC have all convened hearings on automated trading, pointing to likely action on this front soon.
Of these, the Senate Banking Committee—or more specifically its Subcommittee on Securities, Insurance, and Investment—has been the least active to date.22 It held hearings on September 20 and December 18 of 2012 on the topic of the appropriate “Rules of the Road” for computerized trading venues.23 While the nominal topic addressed computerized trading generally and appeared to focus on exchanges and alternative trading systems (ATSs), written testimony from panelists such as David Lauer of Better Markets, Inc.,24 Larry Tabb of the TABB Group,25 Eric Noll of NASDAQ OMX,26 and Robert C. Gasser of Investment Technology Group27 pushed the conversation toward HFT. It remains to be seen how active the subcommittee will be under its new Chairman, Senator Jon Tester.
The SEC, by contrast, has aggressively moved to deal with HFT and computerized trading concerns along six separate lines. First, it has looked to limit the potential impact of runaway algorithms by implementing single-stock circuit breakers that stop trading after large moves and developing the more intricate “Limit Up-Limit Down Plan,” the first phase of which became effective this April 8.28 Second, it convened a roundtable meeting in October 2012 to collect views on how computerized trading companies could reduce the frequency and impact of errors.29 Third, the SEC has adopted the Market Access Rule, Rule 15c3-5, which James Burns, Deputy Director of the Division of Trading and Markets, says “requires broker-dealers to implement controls and supervisory procedures to manage the financial and regulatory risks of market access, including the risks of errant trading algorithms that could seriously disrupt the markets.”30 Fourth, the SEC is actively collecting more data in an effort to understand in near real-time what automated and high-frequency traders are doing in the markets. That effort involves both the planned Consolidated Audit Trail31 and the Market Information Data Analytics System, or MIDAS, which “captures all orders posted on the national exchanges, all modification and cancellation of those orders, all trade execution of those orders, and all off-exchange executions.”32 Fifth, “to use all of these new data tools more effectively, the SEC has hired additional personnel throughout the agency with specialized quantitative and trading expertise.”33 Those “quants” have already begun conducting quant audits, looking at the code and procedures of quantitative trading companies, and the arrival of MIDAS promises to provide them with yet more areas for investigation.34 They have reportedly also recently begun teaming with the Federal Bureau of Investigations “to tackle the threat of market manipulation posed by sophisticated computer trading strategies.”35
Sixth, in a potential sign of things to come, the SEC proposed Regulation SCI on March 7, 2013.36 Building off of its October 2012 roundtable, the SEC proposed rules to govern electronic systems at “SCI Entities,” defined as securities exchanges, ATSs with significant trading volume, and a few other entities.37 The SEC release proposing the regulation highlights “quality standards, testing, and improved error response mechanisms” as areas “needing very thoughtful and focused attention in today’s securities markets”38 and accordingly proposes rules that would require additional testing, business continuity plans, written policies and procedures, and monitoring to keep those policies current.39 Proposed Regulation SCI would also impose new disclosure obligations on SCI Entities.40
The CFTC cannot yet point to as many overt changes in its regulatory regime as a result of concerns regarding automated trading, but it has been very active in collecting opinions on the subject to guide it as it inevitably drafts regulations in this area. On February 9, 2012, the CFTC established a Subcommittee on Automated and High Frequency Trading and four corresponding Working Groups.41 The Working Groups reported to the Subcommittee on March 29, June 20, and October 30 of 2012,42 and the Subcommittee is scheduled to meet again in June 2013. With the tasks of the Working Groups now officially complete, one would expect the Subcommittee soon to report to the CFTC, likely recommending some new regulations.
Areas of Likely Regulation
The evidence presented to the Senate Banking Committee, SEC, and CFTC, along with the questions raised by those bodies in response, points toward areas of particular concern to the government and hence likely areas for future regulation or enhanced enforcement of existing rules. In particular, the following topics have come up frequently:
- Poor IT quality control. While no quality control mechanism can completely eliminate errors, the SEC has indicated an intention to check for what it considers insufficient precaution. As previously reported by Bloomberg, then-Director of the SEC’s Division of Enforcement Robert Khuzami said to expect “a variety of cases” relating to perceived weak controls.43 Former SEC Chairman Mary Schapiro also dismissively referred to the origins of the Facebook and BATS IPO problems as “basic technology 101 issues,”44 phrasing that hints at an expectation of stronger preventative measures in the future.
- Kill switches. Much of the SEC roundtable in October 2012 was dedicated to discussing the minutiae of who should operate these switches, what should trigger them, what their effect should be, and so on.45 Notably, Proposed Regulation SCI does not mandate kill switches, but the issuing release explicitly calls for comments on whether they should be required going forward.46 Unless (as discussed below) the industry takes the lead in establishing standards for such switches, regulation on this topic seems especially likely.
- Excess cancellations. Cancel-to-fill ratios on securities exchanges have been observed to rise as trading speed has increased.47 Both quote stuffing, discussed above, and the generation of flash orders that are intended merely to measure interest at various levels without actually transferring risk48 could contribute to that rise. Regardless of the reason behind growing cancellation frequency, some panel participants have complained about the economic externality that comes when exchanges do not allocate the costs of their networks according to usage, thereby providing firms with an incentive to overtax the network through costless cancellations.49 Hence, both to promote efficient asset allocation and to discourage abusive practices, additional fees for cancellations may well be on their way, whether from the government or from exchanges.
- Proliferation of order types. Recent years have brought a proliferation of new order types, and HFT shops have been the most notable users of some of the more esoteric varieties. Some Senate Banking Committee panel members suggested that such order types do not benefit long-term investors or promote capital raising, risk allocation, price discovery, or any other core purposes of markets.50 Conceivably regulators or exchanges could clamp down on new order types or even eliminate some current ones in response.
- Market fragmentation. Like order types, exchanges themselves have proliferated, now totaling at least thirteen for equities, along with over fifty ATSs.51 Larry Tabb, Founder and CEO of the TABB Group, argued in testimony to the Senate Banking Committee that the complex path that orders follow as a result of this proliferation combined with Regulation NMS allows other market participants, notably HFT companies, to learn details of large trades before they are executed and thus profit by getting ahead of those trades. Whether regulators can prevent such information leakage is unclear, but this seems like an area for potential governmental focus.
- Fragility. Given Regulation NMS, if any national securities exchange has a problem, it affects the whole market system. Similarly, one error by a trading company can have knock-on effects on other market participants.52 At the very least, this provides additional motivation for regulators to require improved quality control within individual market participants, as discussed above. In addition, however, this concern suggests the need for more systemic solutions, such as ways to recognize and quarantine problematic algorithms before they lead to wider disruption.
- Need for data. With the implementation of MIDAS and progress toward the Consolidated Audit Trail, the SEC has made significant strides toward understanding what is happening in the market in or near real time. Still, various proposed controls would require additional information. For instance, kill switches may need to take into account all the trades placed by a particular actor across multiple exchanges and product types. Reflecting such concerns, the CFTC hearings featured discussion of potential additional identification requirements for market participants and automated trading systems.
- Moral hazard from trade breaks. When securities exhibit particularly high volatility that can be traced to an error or an attempt to manipulate the market, exchanges will occasionally bust trades so as to prevent harm to the innocent. The problem with this arrangement is that if market participants know the decision rules exchanges apply to decide when to nullify trades, they can game the system. The difficulty of separating the good actors from the bad has led some53 to call for an end (or at least a significant change) to the practice of busting trades.
- For-profit exchanges. Exchanges have traditionally enjoyed some of the protections afforded self-regulatory organizations as a result of the quasi-governmental role they play in serving the overall market system.54 As exchanges have switched to a for-profit model, some observers, including panel members at the second Banking Committee hearing, have questioned whether such protections are still appropriate. This debate could go either way and is worth watching.
- Minimum price variation. Decimalization in 2001 greatly reduced the tick size for equities, now set by SEC Rule 612 at $0.01 for stocks trading at or above $1 per share and $0.0001 for lower-priced stocks. As required by the Jumpstart Our Business Startups Act (JOBS Act), the SEC issued a report on the effects of decimalization,55 suggesting that larger tick sizes could be appropriate for stocks with small market capitalization, large share price, or low trade volume. In February 2013, the SEC held a roundtable on this issue, including discussion of a potential pilot program.56 While HFT concerns did not motivate the JOBS Act’s requirement for a study, HFT shops are definitely affected by the minimum price variation, insomuch as it affects their ability to improve market prices and get ahead of large orders. A change in tick size could substantially alter the incentive to pursue HFT and thus indirectly market liquidity. Moreover, the same logic that is driving reconsideration of equity tick sizes (such as the trade-off between narrower spreads for investors and corresponding decreases in the incentives for market makers to provide liquidity) could apply equally to other products, suggesting that this is an area for trading companies of all sorts to monitor.
While the regulatory environment surrounding automated and high-frequency trading remains in flux, many industry representatives are proactively taking action: implementing practices that reduce operational and legal risk, coordinating with others in the industry to establish standards that mitigate the need for regulation, and sending comments to Congress and agencies in an attempt to steer the conversation in a direction that addresses investors’ legitimate concerns without adding too many new monitoring costs.
In relation to this industry response, the most important initial observation is that it is not just HFT. Former SEC Chairman Schapiro’s reference to “basic technology 101 issues” could apply equally to all market participants that utilize at least some computerized processes. If kill switches are required, this will likely affect all automated trading companies, and potentially also broker-dealers, exchanges, and technology service providers. Fees for excess cancellations would hit HFT shops the hardest but would also have some marginal effect on the strategy of other trading companies that are seeking narrow edges. If any existing order types are eliminated, this will affect trading by all parties who currently use them, regardless of the frequency of their trading. Similar arguments could be made for several more of the pressure points discussed above, showing that a wide range of market participants have reason to care about regulatory changes that were nominally triggered by concerns over automated and high-frequency trading.
For that matter, there is no universally agreed definition for HFT. The CFTC’s HFT Subcommittee commissioned a working group to develop such a definition,57 but that definition has not yet been broadly adopted. Indeed, the most recent HFT Subcommittee meeting included discussion of triggering regulation not based on satisfaction of a hard-and-fast definition of HFT, which might invite regulatory arbitrage, but rather based on pursuit of trading activity that the regulator would like to discourage regardless of speed. Hence, many companies that would not consider themselves “HFT shops” could get caught in the resulting regulations.
Taken together, these points show how regulations nominally inspired by automated and high-frequency trading could actually affect a much wider array of market participants. And even those companies not directly affected by any of the foregoing points would likely notice the drop in liquidity if new rules cut back on HFT enough, or a tightening of the rules exchanges place on the sorts of algorithms that may submit orders. Such companies thus have at least an indirect reason to follow and comment on developments in this area.
Given the broad potential impact of such new rules, various industry groups are developing coordinated responses to concerns over automated and high-frequency trading. Three prominent industry consortia involved in these efforts are the Futures Industry Association (FIA), the Chicago Quantitative Alliance (CQA), and the AT 9000 initiative (from Automated Trading + ISO 9000).
The FIA has generated and made public several best practices documents.58 As the organization’s name suggests, these documents are primarily geared toward participants in the futures markets, but because they are strategy-independent, they can apply equally to automated, high-frequency, algorithmic, and manual traders. These documents tend to be very detailed but phrased primarily as recommendations rather than requirements.
The CQA59 is currently assembling a best practices document, with a partial draft circulated for review in April 2013. This document is shorter than the combination of the various FIA materials, as it intentionally focuses on only those issues that are specific to quantitative processes. Given how many trading firms now rely on quantitative models for at least some part of their investment process, however, this is still likely to be relevant to many market participants. The goal of this document, broadly speaking, is to share the lessons of past mistakes and of successful companies to establish new, industry-led guidance for quantitative trading. Topics include model validation and testing, internal controls, and compliance rules. Like the FIA materials, this document—by design—goes beyond what current regulations require and recommends various actions, with the understanding that some such actions would be inappropriate for certain firms, such as small trading shops.
The AT 9000 initiative60 approaches this problem from a different angle. Companies in many other industries have implemented variations of the ISO 9000 quality management system standard61 and thereby reduced the frequency of defects and improved customer satisfaction with their products. The AT 9000 project, currently operating under the auspices of Accredited Standards Committee X9, Inc. and targeting adoption by the American National Standards Institute (ANSI) later this year, aims to develop an ISO 9000-style quality management system specifically geared for trading companies with partially or completely automated processes and the various firms that service them. (Once again, given the growing number of firms that rely on at least some automated process, this standard speaks to a large audience.) While it would be impossible to prevent all errors,62 the intention is to reduce the frequency and impact of errors through conscious awareness and continual improvement of a company’s processes.
The shift from a best-practices paradigm to ISO 9000 distinguishes AT 9000 from the FIA and CQA efforts in two key respects.63 First, because AT 9000 is intended to apply to all firms that touch on automated trading, it “travels light,” issuing relatively few recommendations for specific procedures and instead primarily focusing on the duty to record, follow, and improve whatever procedures a particular firm has in place. Second, unlike the CQA and FIA documents, which are intended as general guidance, AT 9000 is designed as a fixed standard with which a given company either does or does not comply. For that reason, a greater percentage of AT 9000′s recommendations are framed in mandatory language than is the case with the other two.
These three initiatives also differ in how they interact with anticipated regulations. The FIA and CQA documents anticipate possible mandates from regulators and encourage firms to implement those controls and other, non-mandatory ones that are expected to reduce operational risk regardless of the actions regulators choose to take. AT 9000, by contrast, in keeping with its mandatory phrasing, is intended to obviate the need for certain regulations, or at least to provide firms with a framework in which to design and express their control processes so as to reduce the difficulty and duration of any regulatory audits they may encounter. In anticipation of such potential interplay between AT 9000 and regulation, members of the AT 9000 drafting committee have presented ideas to Commissioner Scott O’Malia and other CFTC officials,64 and the SEC and other agencies have likewise expressed interest in the AT 9000 project.
Concerns over the potential negative impact of automated and high-frequency trading—both when such trading “goes wrong” as a result of errors and when it “goes right” and achieves traders’ intended results at the cost of overall market quality—are driving review by Congress, the SEC, and the CFTC that is likely to lead to enhanced enforcement of existing rules and issuance of new statutory or regulatory controls.
The SEC’s Proposed Regulation SCI represents one major move in that direction, but additional actions seem probable. Outlined above are some of the common concerns motivating agencies in relation to this topic, recent actions taken by regulators and the Senate to respond to these concerns, and some of the areas in which new rules and enforcement initiatives seem likely to emerge.
While the impetus behind these governmental actions starts with automated trading and HFT, the changes are likely to affect a much wider category of market participants. This provides trading companies, exchanges, and other related entities with an incentive to act now in anticipation of governmental action to come. Industry-led initiatives such as the FIA and CQA best practices documents provide advice on internal processes market participants may want to implement already, to reduce operational risk in any event and legal risk if regulators end up imposing corresponding requirements.
The AT 9000 project aims to simplify the regulations that trading firms and other entities eventually need to follow by proposing a uniform industry standard that is designed to address salient concerns regarding automated trading and to promote continual improvement of companies’ processes. Interested companies can participate in the drafting of the AT 9000 standard today and pursue implementation of the standard once it is finalized and adopted by ANSI.
Mr. Ziliak, email@example.com, has seen investment managers from both sides, working first in-house as a financial engineer and quantitative trader and now as outside counsel representing trading companies. He is a member of the CQA Best Practices and AT 9000 committees discussed in this article.
© 2013 Bloomberg Finance L.P. All rights reserved. Bloomberg Law Reports ® is a registered trademark and service mark of Bloomberg Finance L.P.
This document and any discussions set forth herein are for informational purposes only, and should not be construed as legal advice, which has to be addressed to particular facts and circumstances involved in any given situation. Review or use of the document and any discussions does not create an attorney-client relationship with the author or publisher. To the extent that this document may contain suggested provisions, they will require modification to suit a particular transaction, jurisdiction or situation. Please consult with an attorney with the appropriate level of experience if you have any questions. Any tax information contained in the document or discussions is not intended to be used, and cannot be used, for purposes of avoiding penalties imposed under the United States Internal Revenue Code. Any opinions expressed are those of the author. Bloomberg Finance L.P. and its affiliated entities do not take responsibility for the content in this document or discussions and do not make any representation or warranty as to their completeness or accuracy.