Statistical physics predicts stock market gloom 11:59 02 December 02 NewScientist.com news service
A statistical physics model is predicting that the US stock market recovery suggested by recent rises will only last until spring next year, before tumbling yet further. Physicists Didier Sornette and Wei-Xing Zhou at the University of California in Los Angeles claim to have identified an "anti-bubble" in the Standard and Poor's 500 stock market index. Their model also describes a similar anti-bubble in the Japanese Nikkei index in the early 1990s, which preceded a decade of decline. However, Neil Shephard, an economist at the University of Oxford, UK, is sceptical. "Firstly, the track record of empirical prediction isn't very good and secondly, economic theory says it shouldn't work," he told New Scientist. This is because traders act on new information about the market by buying or selling shares, making it impossible to make a prediction without it affecting the outcome. But the physicists' predictions are in line with those of some others. Haydn Carrington, a dealer at spread betting firm City Index in London, also believes the US market is in a long decline, but that a short term rally is likely: "The Americans are optimistic about recovery, so that will probably happen." Herding behaviour Bubbles and anti-bubbles are traits of herding and imitative behaviour, Sornette says. Investors and traders constantly exchange opinions and information, generating a feedback loop that can drive the performance of the market. A bubble, or bull market, occurs when optimism spreads, pushing the market value artificially high. The bubble may then burst in a dramatic crash, but if not, a slow period of downwards adjustment will follow - a bear market, which Sornette calls the anti-bubble phase. An anti-bubble market has two key characteristics. The value slides inevitably downwards, but oscillates as it does so. The value of the S&P 500 has been riding this rollercoaster since August 2000. Sornette says that the "up" seen now is just one of the oscillations, and that hopes of a recovery will be dashed by a "down" in mid 2003. And the trough that it sinks into may be deeper than this year's low, he says. Failure mechanisms The model used to make this prediction describes "crowd" behaviour of the type Sornette expects from traders and investors. It consists of a set of three equations that describe feedback processes. He developed the equations when studying failure mechanisms in materials - the way that cracks develop and cause damage is similar to the way that information seeps through the market and changes opinion, he believes. The model requires the input of two constants: one quantifies the overall trend (down in an anti-bubble), the other the frequency of oscillation. He chose constants such that the model matched the S&P data from the past few years - and then extended the model to 2004. Journal reference: Quantitative Finance (Vol 2, issue 6) Jenny Hogan
