Thursday, September 12, 2013

Mad, fast and dumb

Recent research from the University of Miami warns that the emergence of the “ultrafast machine ecology,” – referring to algorithmic trading - may create more market crashes. This “predatory” behavior, the author argues, is reaching a stage of instability and may need to be regulated. These arguments, based mostly on an engineering view of financial markets seem to be suffering from a lack of understanding of what markets are. Assuming that regulation is the only medicine for any symptom is problematic for many reasons.

First, let’s look at what financial markets are for. They are set up for the trading of standardized paper that represent the value of real assets. Nobody is forced to buy or sell such instruments and markets fundamentally mean that anybody is able to buy and sell those at any time and ideally at any place. The rules that govern such markets are simple – everybody is allowed to analyze publicly available information and make decisions for herself. Granted, the simple rules – equal access to information, no insider trading and no preferential access to trading are not consistently implemented – but that is a different issue.

Second, application of technology to trade is an innovation – but such innovation has not resulted in risk adjusted excess returns (alpha). Nobody has cornered all the wealth in the world yet by algorithmic trading. So in spite of the mad, fast and dumb bunch that shout from the TV screens every evening, the many supercomputers and light years of cable wrapping around the exchanges and the brilliant physicists and mathematicians who fine tune the algorithms continuously – academic studies show that none of these create alpha. Finally, the presumption that “regulation” will solve these “problems” implicitly assumes that the regulators have perfect information to make the optimal rules. If the last decade is any indication, one cannot imagine regulators are that smart.

Analysis of financial markets and trading behavior based on engineering rules is unlikely to provide any insights. Further, assuming that regulation is needed over the application of emerging technologies for trading is wrong, philosophically and practically. It would be a lot better if regulators focus on implementing the rules that already exist consistently and sending the offenders to jail quickly.