Evolutionary financial market models

We study computer simulations of two financial market models, the second a simplified model of the first. The first is a model of the self-organized formation and breakup of crowds of traders, motivated by the dynamics of competitive evolving systems which shows interesting self-organized critical (SOC)-type behaviour without any fine tuning of control parameters. This SOC-type avalanching and stasis appear as realistic volatility clustering in the price returns time series. The market becomes highly ordered at `crashes' but gradually loses this order through randomization during the intervening stasis periods. The second model is a model of stocks interacting through a competitive evolutionary dynamic in a common stock exchange. This model shows a self-organized `market-confidence'. When this is high the market is stable but when it gets low the market may become highly volatile. Volatile bursts rapidly increase the market confidence again. This model shows a phase transition as temperature parameter is varied. The price returns time series in the transition region is very realistic power-law truncated Levy distribution with clustered volatility and volatility superdiffusion. This model also shows generally positive stock cross-correlations as is observed in real markets. This model may shed some light on why such phenomena are observed.