基于元胞自动机的人工金融市场及其仿真研究

该文通过对金融市场复杂性的分析，并基于元胞自动机和争当少数者模型提出了一个开放的金融预测模型。模型中的投资者相当于争当少数者模型里的agent，每个投资者都必须从初始时各自定义的策略集合中选择最成功的作为每一步的预测策略。同样，投资者也相仿于元胞自动机中的元胞，而选定的预测策略就作为元胞的局部规则，每个元胞都根据局部信息和局部规则做出估价。模型的最终预测值就是由数千个这样的投资者决定的。此模型是开放性的，因为可以通过扩充或重构其预测策略库来获得更精确的预测结果。数值实验显示，虽然策略库比较简单，但其预测的平均相对误差仅为1．73％。

An Artificial Financial Market Based on Cellular Automata and Its Simulation

This paper proposes an open model for finance forecast based on Cellular Automata, Minority Game and the complexity analysis of finance market. The investors in model are regarded as the agents in Minority Game which should select the most successful forecast strategy every simulation time step from the strategy sets defined at the beginning. And also they are modeled as cells in Cellular Automate. Each of the cells works out estimate price with local information and makes the selected strategy as its local rule. The final forecast price of model is determined by thousands of such investors. The model is open since its forecast strategy library can be extended or reconstructed to achieve more accurate forecast results. The numerical experiment results show that the mean forecast relative error is 1.73% even with a simple strategies library.