Beat The Market

Abstract

Speculation in asset market is modelled as a stochastic betting game played by finite number of players and repeated infinite times. With stochastic asset return and unkown quality of public signal, a generic adaptive learning rule is proposed and the corresponding evolutionary dynamics is analyzed. The impact of historical events on players' belief decays over time. It is proved to be a robust approach to adapt to stochastic regime shifts in the market. The market dynamics has characteristics, i.e. endogenous boom-bust cycle, positive correlation in return and volume, and negative first order autocorrelation in return series, commonly observed in financial market but inexplicable by conventional rational expectations theory.