David Pennock of Yahoo Research
Information and complexity in securities markets
Securities markets (e.g., stocks, options, insurance, ... even sports bets) are known to provide accurate forecasts of future events. Recently, a number of new markets and market games have appeared on the Internet with the goal of predicting everything from political elections to Hollywood box office receipts to the discovery of extraterrestrial life. I will describe empirical
studies of these markets, reporting their behavior in terms of information dynamics and (probabilistic) forecast accuracy, including some theoretical underpinnings. I will discuss some of the relevant computational issues, including combinatorics, computational complexity, automated market makers, Bayes-net structured markets, and compound markets based on propositional logic. I will present work on characterizing the nature and limits
of distributed computation in markets.