High Frequency Trading Review


    By Ricky Treadwell

    J. Doyne Farmer is an American physicist and entrepreneur, with interests in chaos theory, complexity and econophysics. He is currrently a Professor of Mathematics at Oxford University. His illustrious career has involved finding a method to beat the casino at roulette as well as conducting the early research for what later became known as chaos theory.

    His main interest in recent years has been developing quantitative theories for social evolution, in particular for financial markets and the evolution of technologies. He was a founder of Prediction Company, a quantitative trading firm that was acquired by UBS, where he was Chief Scientist from 1991-1999.

    I spoke with Doyne after attending an event at the London School of Economics where the UK Government’s recent Foresight Report into the Future of Computer Trading in Financial Markets was being discussed. During his presentation, he theorised that markets can be looked at as an ecology, better studied with theoretical approaches developed in biology than with those usually used in economics. I asked him why this is the case.

    “The difference between economics and biology is that people can reason ‘de novo’. People are smart so they can reason about the world. Computers don’t do that, computers enact code that might have some strategic reasoning built into it but they are basically just running hard-wired rules. High frequency traders have pressure to keep their code short, there’s a limit to how much they can process, and all they can do is give a quick response to a stimulus. That changes how we should perceive the market and makes dealing with rules of thumb and evolutionary knowledge a more suitable model for information processing by computers in markets than de novo reasoning.

    “HFT in particular works that way. Stimulus-response means something happens, you look it up in a table and then you act on it. Computer trading is a highly evolutionary activity. If you have a strategy that makes money then your capital and trading size tend to grow through time and other people catch on to it and their capital grows through time, and the strategies acquire more market power. If the strategy’s not working it dies out. Firms are explicitly evolutionary. Economists – or anyone wanting to understand computer trading – needs to understand that computer trading requires a different style of analysis than the one that economists are used to using.”

    At the same LSE/Foresight conference, Doyne described price-time order execution as an ‘historical accident’. I asked him why he chose to describe it as such and whether he thought pro-rata would be a better execution method.

    “It’s not clear that people consciously chose price-time auctions over pro-rata auctions, it’s just what happened.  A lot of random things happen in the world and it could have been a different way. That was my point. I don’t think there’s anything particularly special about price-time priority, and the world doesn’t have to be that way.

    “I don’t want to say that we should trade pro-rata now, I just think that it deserves careful study. It would slow markets down and take the advantage away from high frequency traders. Because of this I think that it’s likely to be opposed by the exchanges as they make a lot of money from HFT as it pumps up their trading volume.

    “The disadvantage with pro-rata is that you can game it by placing a bigger order than you really want. The drawback with that of course is that someone might hit your gigantic order so if you are gaming, you’re liable to have your large order unexpectedly executed. But even if people aren’t gaming it, bigger traders are going to get more, which could be considered unfair.”

    A hot topic for debate at the conference was the difficulty that academics and regulators have getting hold of the data that they need. I asked Doyne why this is such a problem.

    “There are concerns about confidentiality. Firms are worried that people will reverse engineer what they do and get an advantage. Exchanges and governments are sensitive to that. But they could make older data available or academics could sign up to a strict confidentiality arrangement.”

    What kind of data is required beyond what is currently available?

    “The key thing that is needed are counterparty identifiers for the transactions. That is essential for identifying what’s really going on in the markets and what’s really driving financial instabilities.

    “To use an analogy, trying to understand the ecology of a market based upon data without counterparty identifiers is like looking at a biological ecology and just seeing that ‘an animal’ ate ‘an animal’. Whereas with this data, we can identify and classify what are the animals and how groups of different types of animals affect each other. We can work out who makes profit from whom, how the ecology of traders shifts through time, and what this does to prices that could potentially generate instabilities.”

    Prof. Doyne Farmer, thank you very much for speaking to HFTReview.com.

    If you have any comments relating to this article please join the debate below.

    Related content

    No related content found

    Leave A Reply