Simulation And Research On Agent-based Artificial Stock Market

The stock market is a complex adaptive system.With the discovering of the stock market abnormal phenomena,many traditional financial theories can't give the right explanations.There are many difficulties in empirical research on stock market based on historical data.The Agent-based artificial stock market is the core research field of computational finance,which provides a new research idea for us to study the microstructure and macroscopic phenomenon of stock market.The main work of this paper is to build an artificial stock market based on the real situation of China,through the method of Agent-based computational finance.Based on the bounded rationality heterogeneous body modeling,we build a continuous double auction artificial stock market using the Java language,according to the actual situation of China's stock market,and we also provide a very excellent output interface to support our research.In order to calibration verification of our artificial stock market,first of all,we confirm that China stock market show the peak fat-tailed distribution,volatility clustering features and the long memory features during the past 16 years.After many experiments in our artificial stock market,we examine the simulation data.The result recreates the peak fat-tailed distribution,volatility clustering features and the long memory features.Therefore,our artificial stock market provides a simulation platform for research in trader behavior,design of market rules and information asymmetry.Next,the author design a typical market operation of institutional investor experiment.In the experiment,institutional investors,has amount of money and stock,obtained a huge profit through a typical "wash-dish-cargo-ship" operation behavior,while normal traders suffer great losses.We also test the effect of short selling on the stock market.The results show that the short selling mechanism with margin ratio of 50%will improve market liquidity and stability.Only part of the actual decision-making of traders in read stock market is considered in this paper,so it is not fully reflecting the bounded rationality and heterogeneity of the Agent.The system design is also to be further considered,which is the research direction in the future.