Agent Based Stock Market Simulation And Forecast Research

As artificial stock market as a sub-system of financial markets,it also belongs to the category of complex systems,including a large number of intelligent traders and well defined trade mechanism.due to its inherent complexity of a complex system and the adaptability of smart traders,traditional static mathematical and statistical modeling methods can not meet the development requirements of the current Finance.Since the 1970s,with the rapid development of computer technology and increasingly mature,the financial markets combined with the computer technology which have become a hot research to build a artificial stock market which similar with stock market.The paper based on the artificial financial market simulation pl atform and the calculation principles of economics(ACE).In the sel ection of trading mechanism,continuous double auction mechanism has been selected;The asset structure is consisted of risk assets as stock and risk-free bond;Market traders are divided into fundament al traders and technical traders.It is a big difference with the objec tives the maximization of wealth for most paper.This paper focuses on different asset types to set different investment objectives.We t-ry to design different attributes of agent to fit the real investor about their heterogeneous characteristics.Then compare artificial sto ck data with real data,found that the basic statistical characteristics of price series in the experimental group are similar with real stock price data,including the fat tail,volatility clustering and long mem ory and other data characteristics described by the real stock market.On the basis of the artificial stock market,we will forecast the close price of stock from the real stock market,compared with the prediction of real stock market close price of ARMA model to pro ved that the artificial stock market efficiency;It can provide a refe-rence for investors about investment decision-making in the market.