Study On Agent-Based Stock Market Model

The financial system is generally regarded as a complex system. Agent-basedmodel is the new development of computational finance. Its core idea is thatestablishing a financial system by computer simulation, and using the system for alarge number of simulation experiments to study a variety of complex financialphenomena. This paper is to use agent-based model to simulate the stock market,stock price volatility.This paper first introduces the historical background and development ofcomputational finance. Comparing to traditional methods of mathematical modeling,the Agent-based model uses methods with bottom-up, to relax the assumption ofadvantages. Second, the model of this paper will use the tools of Swarm. The paperwill give a brief introduction of this platform and also its development. And then adetailed introduction of the Agent utility functions will be given. This part of themodel mathematical basis, shows the Agent strategy of selection rules and the law ofthe artificial stock market run. The fourth part named Swarm-JAVA softwareplatform is to establish the steps of the computer simulation of artificial stock marketsystem. The fifth part shows this ASM model’s running results and uses this model todo a series of empirical experiments. This ASM model can simulate the generalcharacteristics of the stock market price fluctuations, the aggregation and so on. Aseries of experiments will also be done to verify that changes in interest rates in thisartificial stock market will affect the stock price, agent different forecasting rules set(total number of rules, learning frequency) will also result different stock situation.The final section summarizes the paper and looks ahead, pointing out what directionof future research can be conducted.