nature.com
Stock market traders show signs of zero intelligence
Economic fluctuations appear to result from random decisions.
24 September 2003
PHILIP BALL
Traders aren't cold rational beings with perfect knowledge.
Market traders are not mindless. But if they were we might not notice the difference, claim J. Doyne Farmer, of the Santa Fe Institute in New Mexico, and co-workers.
Their theoretical model assumes that traders place orders at random rather than on the basis of shrewd calculation and observation of economic trends. It reproduces some of the statistical features of financial markets1. Traders, it suggests, are rather like ants swarming chaotically through the guts of a great clock, barely affecting its ability to tick.
The findings further undermine one of the most cherished notions of economic theory. Since the nineteenth century, economists have imagined traders as omniscient and coldly logical beings, who aim always to maximize their profits based on complete knowledge about what the entire market is doing.
Modern economists recognize that people aren't fully informed or rational. Much recent theory has been devoted to understanding how we cope with imperfect information, and how our behaviour is interdependent and sometimes irrational. Irrationality leads to the herding that can drive market crashes.
All the same, by dispensing with even a restricted form of rationality, the new model is daring. Farmer and colleagues don't imply that anyone really does make decisions by flipping a coin, simply that economic decision-making is so varied and complex that it is hard to distinguish it from random choices.
"The concept of the full-information rational maximizer is one which economics finds very hard to abandon," said economist Paul Ormerod of Volterra Consulting, UK, at this month's conference of the Economic Society of Australia, held in Canberra. "But much of the world is much closer to the random paradigm."
Random work
The model posits two kinds of dealers. Impatient traders submit requests to buy or sell shares immediately at the best available price; these are called market orders. Patient traders specify a worst allowable price; their exchanges are called limit orders. If this price can't be met immediately, limit orders are stored in a queue.
The researchers assume that both types of agent place and cancel orders at random. Each order alters the best prices, affecting the subsequent queuing of limit orders. The statistical properties of prices generated closely match the values seen on the London Stock Exchange for 1998-2000.
This suggests that share prices are set by limit orders altering supply and demand - not by the reasoning behind those orders. Ormerod thinks that this apparent randomness comes from the imperfect ability of market agents to process the complex and often incomplete information at their disposal. The best strategy is virtually impossible to find, so everyone reaches different conclusions.
References
Farmer, J. D., Patelli, P. & Zovko, I. I. The predictive power of zero intelligence in financial markets. Preprint, xxx.arxiv.org, (2003). |Article|
© Nature News Service / Macmillan Magazines Ltd 2003 |
