The Simulation Of The Statistical Character Of High Frequency Data Of Chinese Stock Market With Voter Model

In this paper, we apply the voter model which come from the interacting particle system to the stock market. We construct a new model to simulate the return process of stocks. The paper contains two parts:In Part One, we analyze the statistical properties of the high-frequency data in stock market. Up to November 17th, 2006, the capability of Chinese stock market reaches 6.51 trillion, it's proportion in Chinese gross domestic product is more than 46%. The force of Chinese stock market to Chinese economic is better and better, it gets the attention of the people study theory and practice. So we analyze the statistical properties of the high-frequency data in stock market, we hope to find some statistical rule of these data. We use the nominal distribution paper of statistics and the methods of statistical physics, such as the log-log plot,the power law distribution test and Zipf plot. Using the Matlab 6.5,SPSS12.0 and EVIEWS5.0, we get the fat-tail phenomena and partial character of the return process of Chinese stock market. We also analyze the correlation among the stock markets of China,America and Hongkong.In pert two, we construct the return process using voter model. From the conclusion of part one, we know that the traditional normal distribution hypothesis is not right. The particle system including voter model comes of 1960s, it's mainly used to dealing with how a system react for the information and the result after reaction. We try to apply the voter model to dealing with the infection of information to stock price. We mainly talk about how to forecast the stock price through the reaction of the information to the return process.We find that the statistical property of voter model is very like that of the return process of Chinese stock market from computer simulation, and the result of simulation by voter model is much better than normal distribution.