One Variable Too High
For years, researchers have been looking for increasingly accurate ways to use computers to model market activity. Now comes an unexpected breakthrough:
A model that assumes stock market traders have zero intelligence has been found to mimic the behaviour of the London Stock Exchange very closely.
[Insert Cheap Shot Here.]
The article goes on to explain that this doesn't mean that people are out there trading randomly (the way the model does) but rather that a model that makes random trades better reflects the price variance and volatility of the market than does one that attempts to take human intelligence into consideration. At this point, it could just be that the intelligence of human agents in the system is too complex (or too subtle) to be modeled accurately.
While a model with zero-intelligence traders may reflect certain characteristics of the real market accurately, one that I would venture to guess is missing is the intelligence of the market itself. As TradeSports, the Foresight Exchange, and the Iowa Electronic Markets have consitently demonstrated, markets are better at spotting trends and predicting outcomes than individual experts could ever hope to be. Which raises an interesting question: If one were to create a zero-intelligence-trader model of one of those markets, would it behave like that market?
I imagine it would, at least where price variance and volatility are concerned. But as a predictive tool? It would be utterly worthless. The intelligence of the market is apparently an emergent property. Viewed as individual transactions, the model with mindless traders and the actual market where people are looking to make the best predictions possible, the two markets would look about the same. But test each market against it's ability to actually predict outcomes — it won't even be close.