CFTC Working Group Tries to Define High Frequency Trading
|Jul 5th, 2012 | Filed under: Algorithmic and high-frequency trading, Commodities, Today's Post | By: cfaille||
The Technology Advisory Committee of the Commodity Futures Trading Commission met on June 20 to discuss, among other matters, the proper definition for high frequency trading.
They were there to hear from four working groups. The first of these – the one that had been laboring on that definition – had the driest sounding purpose, but it ended up proving a good morning’s excitement.
Lions have Prides, Cheetahs don’t
But first, there were opening statements to deliver. Commissioner Scott O’Malia, the chairman of the TAC and the driving force behind the CFTC’s determination to do something about these threatening speeds, had little to say except to congratulate committee and working group members on their hard work. He also mentioned that Chief Economic Andrei Kirilenko is leaving soon. Kirilenko is going to be teaching at MIT starting next January.
Gary Gensler, the chairman of the CFTC, said that the CFTC has to adapt to the reality that trading is moving from human decisions to those of machines, and he said that the CFTC is at work on a concept release “on potential risk controls and system safeguards for electronic trading platforms.”
Another Commissioner, Bart Chilton, then spoke about “cheetahs,” his nickname for high-speed traders. He believes cheetahs ought to be registered and on the CFTC’s radar screen. Later in the proceedings, he would say that the executives to whom these traders report, the “head of the prides” of the cheetahs, ought to be responsible too.
Once this first working group was seated, Greg Wood, of Deutsche Bank Securities, took the lead in its presentation.
So here’s the draft definition: HFT is a form of automated trading that employs: (a) algorithms for decision making, order initiation, generation, routing, or execution, for each individual transaction without human direction; (b) low-latency technology that is designed to minimize response times, including proximity and co-location services; (c) high-speed connections to markets for order entry; and (d) high message rates (orders, quotes, or cancellations).
Wood emphasized that it was a deliberate choice of the working group that this definition doesn’t mention holding times, or “good or bad behaviors,” such as the provision of liquidity to the market on the one hand or market abuse on the other.
It became clear during the presentation, though, that the group was not all of one mind on this point. One of the members, Joseph Saluzzi, the co-head of equity trading at Themis Trading LLC, an agency brokerage, was especially outspoken about his disagreements.
Saluzzi is also the co-author (with his Themis Trading colleague Sal Arnuk) of a recent book on the subject, Broken Markets. Saluzzi is firmly of the view that HFT and related features of the structure of contemporary markets have undermined the capital raising function of exchanges, turning them into casinos.
But Arnuk’s and Saluzzi’s focus is precisely on holding times. The problem in their eyes with HFT is not that a purchase can take place very quickly, with little latency. The problem, rather, is that the purchase is almost immediately followed by a sale; a position is created only to be exited. On Saluzzi’s view, any definition of HFT that misses this misses the point of worrying about HFT.
At the CFTC hearing, Saluzzi cited a paper co-authored by the CFTC’s own economist, Kirilenko, saying that high frequency traders “aggressively take liquidity from the market when prices were about to change and actively keep inventories near a target inventory level.” Those are notions that, he believes, ought to be incorporated into the CFTC’s definition.
Other Sticky Wickets
The reference to “proximity” in part (b) of the above definition also seems to have generated a good deal of discussion within the working group.
Working “proximity” within the definition, Wood mentions, could create opportunities for arbitrage, had it been done too narrowly. But the wording avoids this problem, and acknowledges the reality that if HFT is the objective, then the “algorithm should be as close as possible to the marketplace,” depending on what is available and depending on the trading strategy involved.
Another question: why does the definition not mention data consumption? Speed is defined in terms of speed of order entry. But what about the speed with which an algorithm can aggregate news and other data? “These are things that we feel are still left open,” he said.
The bottom line, still, is that the working group, or most of it, believes that it has provided the TAC and the CFTC with a working definition that can serve as a basis for further discussions and for market surveillance.
One group member, also a CFTC staffer, Joan Manley, discussed the use of the word “transaction” in the first prong of the definition. A “transaction” is a very broad term at law, she emphasized, broader than “contract” or “agreement.”
One key point about this language is that human decisions replace algorithms, and move trading outside of the realm of this definition, only when “each” transaction depends upon a human decision. A firm could not easily continue the pattern of its trading, its strategy for getting alpha, simply by having some human do some sporadic pointing and clicking with a mouse. “You’d be doing this all day,” she said, tapping the desk quickly with her finger.
Christopher Faille is a Jamesian pragmatist. William James has taught him, for example, that "you can say of a line that it runs east, or you can say that it runs west, and the line per se accepts both descriptions without rebelling at the inconsistency."