Market Algorithms Got It Wrong
Read in Le Monde newspaper (yes I know, nobody’s perfect), the SEC (Securities and Exchange Commission) and the CFTC (Commodity Futures Trading Commission) published a couple of days ago their final report on what happened on May 6, 2010.
I was in front of the Bloomberg screen, what I saw was the stuff of legend: In a nutshell, the products I was supposed to check on do not usually observe high levels of volatility in a trading day, so that was quite disturbing to see the S&P 500 volatility index (VIX index) leap up and down (while the S&P and Dow Jones went down for an hour or so).
There was a 15mins delay, but it was quite impressive. In facts the SEC report describes it in precise terms: ” May 6 started as an unusually turbulent day for the markets. […] At about 1 p.m., the Euro began a sharp decline against both the U.S Dollar and Japanese Yen.
Around 1:00 p.m., broadly negative market sentiment was already affecting an increase in the price volatility of some individual securities. […] Equities listed and traded […] began to substantially increase above average levels. By 2:30 p.m., the S&P 500 volatility index (“VIX”) was up 22.5 percent from the opening level […] the E-Mini S&P 500 futures contracts (the “E-Mini”), as well as the S&P 500 SPDR exchange traded fund […] had fallen from the early-morning level of nearly $6 billion dollars to $2.65 billion (representing a 55% decline) […] against this backdrop of unusually high volatility and thinning liquidity, a large fundamental5 trader (a mutual fund complex) initiated a sell program […] as a hedge to an existing equity position.This trader chose to execute this sell program via an automated execution algorithm (“Sell Algorithm”) that was programmed to […]target an execution rate set to 9% of the trading volume calculated over the previous minute, but without regard to price or time.
However, on May 6, when markets were already under stress, the Sell Algorithm chosen by the large trader to only target trading volume, and neither price nor time, executed the sell program extremely rapidly in just 20 minutes“.
So there we go, nobody’s fault, the Algorithm just went berserk and that’s it. No, things are more complicated and even more bothering than it was made to look. It apparently just went by unnoticed. I mean journalists got us used to the fact that everyday is doomsday, that if it was not financial apocalypse, it is World War III or Judgement day. On that particular instance, the Financial Times did not feature a particularly concerned tone. In facts, they dismissed “quote-stuffing” as the main reason for the brief but high volatility on the markets. In broad terms, quote-stuffing is operated through a trading program (a routine macro) that ensures to get every arbitrage opportunity out of tiny price differential and in a very brief amount of time (no more than a dozen seconds).
There was however a very sensible article in the Financial Times that provides just the kind of balanced insight on the matter; The idea is than when financial markets are unstable, very volatile, an algorithm program that prices routinely securities tend to amplify the trend; a time-out would be of some benefit, but again, the article shies away from the issue: how are these algorithm computed, and could regulatory bodies such as the SEC in the US capable of monitoring future anomalous trades before in order to avert further damages? To the first question, that should be up to the users (the traders) and the quant people that design them. It must be pointed out that the models are so complex that at times, one could believe they are able to stick closely to actual figures and real numbers. But because they are mathematical, there are bits of informations that are lost in the process, and the more derivative these securities are, the less information available is, not to mention the underlying assumptions behind even the most basic model, that are not always -if never- verified in actual world. It is true the past financial meltdown prompted some financial players to review their models, but when one gets use to an alluring mathematical model, and when everyone gets back “business as usual” there is little incentive to change. The regulators, on the other hand, have long since lost any real power to stop or monitor abnormal trades.
IT has taken over and regulators can do little about it. It is quite strange that financial markets, that are considered to be the most pure form of perfect and efficient market structure act at times (and increasingly so) completely free of regulation or institutional framework. It is as though the basic rules of the game were enough, and because players have in some ways internalize other institutional rules, SEC and others cannot act in pre-emptive manner, nor swiftly enough to limit the damage. And that’s how a sub-optimal equilibrium in Game Theory is reached with perfectly rational players; Algorithms are very useful indeed, but are they so without human supervision?
PS: FT registration is free but compulsory to read the articles.