FIA special report: CFTC discusses high-frequency trading

The Commodity Futures Trading Commission hosted a lengthy discussion of high-frequency trading at a public meeting of its Technology Advisory Committee on June 20. The meeting was organized by CFTC Commissioner Scott O’Malia, the chairman of the TAC, and served as a forum for the TAC to review a proposed definition of HFT drafted by a working group of industry experts that includes several representatives of FIA member firms. The TAC also discussed a range of related issues, including different types of HFT strategies and methods for tracking their activities.

It is not clear what effect the discussion will have on the CFTC’s rulemaking process. While there appeared to be a consensus among the industry experts on certain technical aspects of the HFT definition, several members of the TAC expressed concerns about how this definition might work in practice and warned against attempting to use this definition as the basis for new registration rules or reporting requirements. In addition, the working group does not expect to finalize the definition until October, and even then it will only be anon-binding recommendation.

O’Malia, a Republican, was at pains to stress that the purpose of the TAC’s examination of HFT was to gather expert advice on market trends rather than to build the foundation for new rules. “I didn’t assemble this group to develop rulemakings,” O’Malia said. “I brought them together to help inform the thinking of the market, the public and the commission about high-frequency trading.”

Nonetheless, O’Malia called for further work on trading strategies associated with HFT to help the agency better understand market practices. In addition, Andrei Kirilenko, the CFTC’s chief economist, commented that the agency is mulling the idea of “tagging” certain types of trading strategies in order to improve its surveillance of market activity.

Commissioner Bart Chilton, a Democrat, gave a more direct indication of what regulations may emerge from this process. In his opening remarks at the meeting, Chilton called for action in four areas: registration of high-frequency traders, testing of algorithms, installation of “kill switches” to stop trading, and quarterly reports on “wash sales” where two trading programs within the same firm are on opposite sides of the same trade.

Meanwhile CFTC staffers are continuing to work on a separate HFT-related initiative under the direction of CFTC Chairman Gary Gensler. During the course of the meeting, Gensler reiterated his plan to publish a “concept release” on technology-driven changes in the markets that the CFTC oversees. Gensler said this concept release will ask for public input on risk controls and “system safeguards” for electronic trading.

Gensler also signaled his concerns about the need to address the risks of automated trading. “Regulators cannot assume that the algorithms in the markets are well designed, tested or supervised,” he warned. “To give hedgers and investors confidence in markets, our regulations have to adapt to markets that are increasingly moving from man to machine.” The TAC discussion on HFT lasted for more than three hours and covered presentations from four working groups. The TAC’s next meeting is scheduled to take place in Chicago during the FIA Expo Oct. 30-Nov. 1.

Highlights of TAC Discussion

Draft Definition

The working group that drafted the HFT definition consisted of two CFTC staffers and six industry representatives: Sean Castette, Getco; Colin Clark, NYSE Euronext; Chris Concannon, Virtu Financial; Joseph Saluzzi, Themis Trading; Larry Tabb, Tabb Group; and Greg Wood, Deutsche Bank Securities.

Wood, who led the presentation to the TAC, explained that the group intentionally chose to focus on the “mechanics” of HFT and avoided taking a position on trading strategies or policy issues such as whether HFT is good or bad. He also stressed that trading systems would not be captured by this definition unless they met all four criteria in the definition.

The working group’s draft definition reads as follows:

High frequency trading is a form of automated trading that employs:

• algorithms for decision making, order initiation, generation, routing, or execution, for each individual transaction without human direction;
• low-latency technology that is designed to minimize response times, including proximity and co-location services;
• high speed connections to markets for order entry; and
• high message rates (orders, quotes or cancellations).

Wood noted that the group did not come up with any specific numbers or thresholds for the fourth criteria on “high message rates” and explained that this will require considering a number of factors, such as the beneficial owner behind the activity, the length of the observation period, the comparison to overall market activity, and the relative concentration or fragmentation of activity. Concannon of Virtu Financial added that thresholds would have to be adjusted to the amount of activity in specific markets. So for example, he explained, in a back month of a futures contract it might only take a few small trades to quality as a “high” level of activity.

Several participants expressed concern that the definition was too broad. For example, Marshall Terry of South Ferry Capital Partners, a small hedge fund, expressed alarm at the possibility that his firm would be captured by the definition if it routes trades to the market through an algorithmic trading tool, especially if the definition is used to establish a new registration requirement. Joe Saluzzi of Themis Trading, an interdealer broker, also expressed alarm, saying many of his institutional investor clients would fit the definition because they use algorithmic trading tools to execute their trades. Saluzzi said the definition should distinguish between good and bad HFT and should not capture large money managers.

This prompted Concannon to comment that market disruption can come from any type of market participant misusing algorithmic trading tools. Wood added that defining HFT more narrowly would increase the potential for regulatory arbitrage. The discussion also prompted one of the CFTC staffers in the working group to comment on how the definition might be used for regulatory purposes. The CFTC staffer said that the regulatory burden would fall on the brokers or vendors who provide the algorithm, rather than the clients who use the algorithm. “Control will be key in the next phase,” the staffer said.

Trading Strategies

The TAC also discussed a proposed classification and “taxonomy” of trading strategies that use HFT. This taxonomy was developed jointly by two working groups consisting of CFTC staffers and industry experts. The taxonomy divided trading strategies into four broad families—arbitrage, short-term directional trading, liquidity provision and liquiditydetection—and a large number of sub-strategies in each family. For example, the “liquidity provision” family included the following five sub-strategies: rebate capture, quote matching, spread capture, market making, and layering. (Click here for proposed taxonomy.)

In contrast to the discussion on the proposed definition, there was much less agreement on the validity and utility of this approach to understanding HFT. The members of the working groups said that it is not always possible to associate an order with a specific strategy because the order may be part of multiple trading strategies and its purpose may change over time. In addition, trading strategies are constantly evolving and may shift from one type to another and back again.

Richard Gorelick of RGM Advisers, a member of the TAC, added that a taxonomy may be useful when distinguishing different species of animals but trading strategies often do not fit into rigid categories. “It’s very clear in the animal kingdom that you’re either a mammal or you’re a bird,” said Gorelick. In contrast, a trading strategy can have elements from more than one category, he said. Like a platypus, an Australian mammal that has a bill like a duck and lays eggs, a trading strategy can be “sometimes a bird, sometimes a mammal.”

Tagging and Surveillance

Another topic of discussion was the oversight and surveillance of HFT. This working group, which like the others included CFTC staffers as well as industry experts, assessed the data already being collected by exchanges and concluded that creating a new “tagging” system for tracking HFT would not be necessary.

Dean Payton of CME Group, who led this group’s presentation, explained that CME, IntercontinentalExchange and National Futures Association, the self-regulatoryorganizations for the U.S. futures industry, have systems in place to identify automated trading systems and measure the volume of trades and order message by firm, account or operator. Attempting to apply “strategy-related identifiers” or registration requirements for algorithms would not be an effective use of regulatory resources, he said.

“We can track the data in very granular way,” Payton told the TAC. “Artificial distinctions do not move us forward in terms of defining where the problems are.” The focus instead should be “on the kinds of conduct that you want to prohibit.” 

Cross-market surveillance was the one area where the working group did see a role for CFTC action, mainly because each SRO only has insights into its own market. This will be especially important when the swaps market migrates to swap execution facilities, several participants commented. That is where the federal government can “add value” by improving its data capture and technology, Payton said.