Computational experiments successfully predict the emergence of autocorrelations in ultra-high-frequency stock returns

Social and economic systems are complex adaptive systems, in which heterogenous agents interact and evolve in a self-organized manner, and macroscopic laws emerge from microscopic properties. To understand the behaviors of complex systems, computational experiments based on physical and mathematical models provide a useful tools. Here, we perform computational experiments using a phenomenological order-driven model called the modified Mike-Farmer (MMF) to predict the impacts of order flows on the autocorrelations in ultra-high-frequency returns, quantified by Hurst index $H_r$. Three possible determinants embedded in the MMF model are investigated, including the Hurst index $H_s$ of order directions, the Hurst index $H_x$ and the power-law tail index $\alpha_x$ of the relative prices of placed orders. The computational experiments predict that $H_r$ 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 $H_r$. The predictions of the MMF model on the dependence of $H_r$ upon $H_s$ and $H_x$ are verified by the empirical results obtained from the order flow data of 43 Chinese stocks.