I’m just returning from a conference on High-Frequency and Algorithmic trading organised by City University of Hong Kong (which has no relation to City University, London). A couple of papers caught my attention.
First, Matthew Baron presented a paper, jointly written with Jonathan Brogaard and Andrei Kirilenko, called “Risk and Return in High Frequency Trading”. This paper uses proprietary CFTC data that allows the researchers to identify high-frequency trading firms and to track their trading activity in the E-mini S&P 500 futures contract. The authors show that the levels of HFT trading revenues and the consistency of those revenues over time are both breathtaking. HFT firms that are aggressive (i.e. they take liquidity from markets rather than making markets) earn excess returns of around 100% per annum and the best performing funds have Sharpe Ratios in the double digits. They proceed to show that the ranking of HFTs in terms of the returns they earn is highly persistent. The best performing HF traders this month will also tend to perform very well next month. This persistence is something we do not often see in other contexts (e.g. mutual fund performance) and suggests that the HFT world might have important differences in industrial organisation terms when compared to other parts of the investment industry.
The second paper that caught my eye was Adam Clark-Joseph’s paper called ‘Exploratory Trading’. This paper outlines a particular strategy that aggressive HFT traders use to generate trading revenues. Again the context is the E-mini S&P future and again the data used comes from the CFTC. His ‘exploratory trading’ strategy goes something like this. HFT firms probe markets using small aggressive orders to work out how much price impact trades are likely to have (i.e. how much the market will move, in current conditions, in response to a trade). These small orders are the ‘exploratory trades’. Then, at times when the exploratory trades indicate that price impacts will be large, HFTs trade in front of the predictable demand that comes from slow traders. Assume, for example, that a slow trader is buying. The HFT, through its exploratory trades, has worked out that the slow buy is going to move prices a long way up. So by buying in front of the slow trader and selling when the slow trader has completed his purchase, the HFT profits from the impact the slow trader has had on the market.
So together, these papers tell you that HFTs make very decent trading revenues and they also give you some indication of how some of them make their money. People have argued that the fact that HFT revenues are so strong indicates that they have an unfair advantage. Others may say that exploratory trading of the type that Clark-Joseph identifies should be outlawed as it hurts slow investors. I don’t subscribe to either of these views. Clark-Joseph’s work tells us that some HFTs use expensive and complex investments in technology to devise clever strategies that allow them to generate superior market intelligence. It’s not obvious to me why we should complain about this. As long as it’s within the rules of the game, these profits are rents for understanding markets and technology and being smart. It’s probably fair to say that this type of HFT does not create much social value, though, but that’s a charge that can be levelled at many activities.