On Monday, the CFTC held a research conference where their chief economist, Andrei Kirilenko, presented the results of a new paper titled “The Trading Profits of High Frequency Traders” (this is the draft copy since the paper has not yet been officially released). This paper has already stirred up quite a bit of controversy as the pro-HFT crowd race to try and debunk it.
First, we would like to say that we have had the pleasure of working with Andrei Kirilenko and found him to be one of the most knowledgeable regulators that we know when it comes to high frequency trading and market structure. He certainly is not a newcomer to the field and has published numerous papers on HFT. He was also one of the members of the Joint CFTC-SEC Advisory Committee which published the Flash Crash report. The man knows what he is talking about.
Now, let’s get to the paper. Before even reading past the second sentence of the abstract, something should strike you that this paper is unique when it comes to academic literature on HFT. The second sentence reads “Using transaction level data with user identifications, we find that high frequency trading (HFT) is highly profitable.” The unique part is not that HFT is profitable (we all know that), it’s that Kirilenko and his fellow authors were able to use data that had “user identifications”. This means they were not guessing, which is what most other academic literature does since it usually relies only a sample set of data that is provided by the exchanges. Kirilenko backed this up yesterday in a NY Times article by saying: “We’re not estimating…Our data is excellent”.
The paper focuses on the level of profits that HFT’s are generating and has four main findings:
“First, HFTs are profitable, especially Aggressive (liquidity-taking) HFTs, and generate high Sharpe ratios.
Second, HFTs generate their profits from all other market participants, and do so mainly in the short and medium run (seconds to minutes).
Third, firm concentration in the HFT industry is not decreasing over time, nor is its profitability.
Fourth, HFTs profits are persistent, new entrants have a higher propensity to underperform and exit, and the fastest firms (in absolute and in relative terms) make up the upper tail of performance. ”
The authors separate HFT’s into three categories: Aggressive, Mixed and Passive. The aggressive HFT’s (those who take the most liquidity) are the most profitable. A bit of a side note here. The paper defines HFT’s as : “HFTs are identified as those firms with high volume, low intraday inventory and low overnight inventory.” As many of you know, position inventory was a contentious point in the CFTC’s recent definition of HFT. We argued that it should be part of the definition but were unable to get it into the final definition . We are happy to see that the inventory component is part of the definition that Kirilenko used for this paper.
Some other facts from the paper:
-HFT represents 54% of the average 3.2 million E-mini contracts traded per day.
-HFT’s limit the positions they take intraday
-Aggressive HFT’s earn an average of $45,267 per day in profits while Mixed and Passive HFT’s earn significantly less ($19,466 and $2,460 per day, respectively). This suggests there is a strong profit motive for liquidity taking rather than liquidity providing.
-Persistent profits over time, a common feature among HFT firms, suggests that something other than luck is driving a firm’s performance.
-New HFT entrants are not at a disadvantage, and may outperform, but that after their initial entry, as soon as three or four months after, they began to underperform.
-The relationship between speed and profits shows that there is an association between the two.
The paper left us wondering: Who are these handful of aggressive HFT’s that consistently make money trading the e-minis? Are they large enough to be doing what is known as “momentum ignition” in the most liquid futures contract in the world? What effect is this type of trading having on other asset classes? Is this type of trading purely momentum and divorced from economic fundamentals?
While we probably won’t get answers to these questions any time soon, we believe this paper is a groundbreaking paper that will change the way academics study HFT. There is, however, one major problem. The paper only analyzes futures data. Most large HFT’s trade in multiple asset classes including stocks, ETF’s, futures, options and currencies. We know that HFT’s make their money differently in the equities market. The concepts of rebates, order types and data feeds are much more important in the equities space.
The Kirilenko paper shines a very bright light on the practice of aggressive HFT trading in the e-mini contract but we are still left to wonder how much money is being made and what is actually going on in all those other asset classes. Since Mr. Kirilenko is the Chief Economist at the CFTC, he was able to obtain very clean data down to the user level for the e-mini futures. Wouldn’t it be nice if the SEC could do the same for the equity and options pieces of the puzzle?
We hope this paper motivates the folks at the SEC (and the stock exchanges) to produce a report with this level of detail. Right now, the Kirilenko report fills one piece of the puzzle in. It’s up to the SEC now to fill in the rest of the pieces.