Computer simulation of the Cont-Bouchaud percolation model of financial markets

Abstract

We propose a simple model that is capable of generating the main ‘stylized facts’ observed in Empirical data: fat tailed probability distribution of logarithms of relative price changes, slow decay of the autocorrelation of absolute value of logarithms of relative price changes, volatility clustering and fast decay of the autocorrelation of logarithms of relative price changes. In this model, traders congragate together a sin the Cont-Bouchaud model to form clusters of different sizes. As in centralized financial market trading takes place through a specialist, the market maker, who is responsible for updating prices for the next trading session. A second specialist, the market regulator, is also incorporated in the model, whose prime responsibility is to limit speculative behavior in the market.