Deja vu and the myopia of our spirit

Amongst the stormy seas of markets recently (off the back of a credit and liquidity crunch apparently initiated by ongoing and deepening problems with sub-prime loans in the US and the related CDOs), bobs the grey and bloated bodies of a clich├ęd failure.

Unwavering belief in trends, normal market conditions and trading rules developed out of a less than infinite history of prices have again resulted in burnt fingers and an abundance of flotsam and jetsam on the high seas of international markets. Computer and algorithm-driven “quant funds” have apparently taken a beating in the “unusual” market conditions of late. These systems are usually calibrated to a period of history, to identify profitable trading strategies based on complicated models, multiple factors and supposedly rigorous statistical analysis.
High volatility and correlation across markets took down LTCM (read When Genius Failed: The Rise and Fall of Long-Term Capital Management) before. So-called “programme-trading” or “portfolio insurance” that was blamed (non incontroversially and not fully substantiated) for the 1987 market crash. Portfolio insurance is still alive and well in the form of delta hedging. Turns out the old name had a rather negative taint to it. Don’t get the wrong idea, I’m not against delta-hedging, or any specific trading strategy. I’m just not convinced that the results are all they’re cracked up to be. A system that works well some of the time then fails spectacularly every now and then is not my idea of a good night’s sleep, or a sustainable long-term strategy.

Goldman Sach’s apparently still believes in the system. Then again, they have to say that, don’t they?

The developers of these systems would do well to look at the past from a human and historical view rather than just a limited slice of a time-series. It’s too easy to consider recent history as representative of the future. We all do it, but the trick is to maintain some scepticism and not get carried away by hope, greed and fear.