Microscopic determinants of the weak-form efficiency of an artificial order-driven stock market

Stock markets are efficient in the weak form in the sense that no significant autocorrelations can be identified in the returns. However, the microscopic mechanisms are unclear. We aim at understanding the impacts of order flows on the weak-form efficiency through computational experiments based on an empirical order-driven model. Three possible determinants embedded in the model are investigated, including the tail heaviness of relative prices of the placed orders characterized by the tail index $\alpha_x$, the degree of long memory in relative prices quantified by its Hurst index $H_x$, and the strength of long memory in order direction depicted by $H_x$. It is found that the degree of autocorrelations in returns (quantified by its Hurst index $H_r$) is negatively correlated with $\alpha_x$ and $H_x$ and positively correlated with $H_s$. In addition, the values of $\alpha_x$ and $H_x$ have negligible impacts on $H_r$, whereas $H_s$ exhibits a dominating impact on $H_r$. Our results suggest that stock markets are complex adaptive systems and self-organize to a critical state in which the returns are not correlated.