Economic Models for Stock Markets Should Incorporate “Outlier” Events
So‑called “outliers,” which are rare events in trading on the New York Stock Exchange (NYSE), actually occur in regular patterns, and thus should be incorporated into economic theories, which to date have dismissed them as “anomalies,” according to Eugene Stanley of Boston University.
Stanley spoke at the APS March Meeting in New Orleans about his efforts to uncover whether there are underlying unifying principles–the equivalent of physical laws–that dominate financial markets.
“Classic economic theories not only fail for a few outliers, but there occur similar outliers of every possible size,” he said. “So ignoring them is not a responsible option.” Stanley has recently completed analysis of 200 million transactions on the NYSE over a two‑year period–significantly more data than has been available for similar analysis previously.
Econophysics emerged in the mid‑1990s when several physicists–including Stanley–decided to apply the tools of statistical mechanics to the complex problems posed by financial markets. Huge amounts of financial data suddenly become available at that time, and there were an increasing number of PhD physicists finding work on Wall Street as financial analysts.
Many different physics models have been applied to financial systems, including percolation models, diffusion theory, models with self‑organizing criticality of complexity, models developed for earthquake prediction, even chaotic models originally developed to study cardiac arrest. Fractal analyses of cardiac rhythms suggest that healthy people have complex cardiac behavior, compared to the rhythms of the unhealthy. Researchers at Brigham Young University are looking into whether similar complexity might be an indication of a healthy company.
Stanley uses a spin glass model to describe stock market fluctuations. The stock market is a complex system made of up many individual units (traders) who interact and make decisions based on the relative strengths of those interactions. The stronger the interaction –the more trustworthy a trader deems a colleague–the more influence that interaction has. But the strength of those interactions can change with time, if for example, a trader loses confidence in a colleague.
Of course, no financial model is likely to ever enable analysts to predict a specific event in the stock market, any more than one can precisely pinpoint the time, location, and severity of an earthquake. One of the prevailing economic theories is the random walk hypothesis for stock market prices holds that prices can’t be predicted due to the lack of correlation between past and present prices. Just because a stock rises one day, there is no guarantee it will rise again the next.
Stanley was unequivocal about this, calling the stock market “a very complex system and probably insoluble,” emphasizing, “There is absolutely no way anyone has been, or will be able to predict the future.” However, better models that take outlier events into account can help investors better manage risk.