Agent-based Stock Market Modeling And Market Fractal Property Research

Since Fractal Market Hypothesis proposed, the study on fractal characteristics of financial market has become a hot topic in academia. Up to now, there are a large number of research results about market fractal characteristics, but most of these works focus on testing the fractal characteristics of kinds of securities asset prices or returns and drawing some meaningful conclusions based on the strength of multi-fractal property from the perspective of empirical testing. There is few works about the causes or influencing factors of time series fractal characteristics. Therefore, based on constructing agent-based stock market model, we research the influencing factors of times series fractal characteristics, moreover, analyze the relationship between investment term structure and market volatility in Fractal Market Hypothesis. Main contributions and innovative achievements are as follows.(1)From the perspective of interaction among agents, we construct a agent-based stock market model, which constitutes four kinds of agents like fundamentalists, momentum, reversal and noise traders and rules corresponding to Chinese A share market. Analyzing and inspecting the model data, we find that the model time series own multi fractal property, the peak and fat-tail property of return distribution, Inverse cubic law, long-term correlation, and the model time series statistic values coincide with empirical results while the model parameters are in a certain range.(2) From the perspective of system integration, we integrate the multi-agent model and financial market analysis algorithm to build a simulation and analysis system with functions of data collection, model simulation and data analysis. With the system, users could easily control the simulation process of multi-agent stock market model, analyze the model data or import data, query and download the securities assets deal data in real market.(3)Based on multi-agent model, we analyze the influencing factors of time series multi-fractal characteristics from the perspective of variation characteristics of returns time series and agents trading behavior, the results prove that the fat-tail property of return distribution and long-term correlation also contribute to the time series multi-fractal property, and compared with fat-tail property of positive return distribution, the fat-tail property of negative return distribution contributes to the time series multi-fractal property more greatly。As the chartist feedback time shortened, the multi-fractal spectrum width of return time series show a rising trend, but the relationship between reversal traders’feedback time and multi-fractal property is not obvious. With the decrease of the fundamentalists’market participation, the multi-fractal spectrum width of return time series is increased obviously. But With the decrease of the reversal traders’market participation, the multi-fractal spectrum width of return time series shows a decreasing trend, the significant of multi-fractal property is weakened.(4)With analyzing the relationship between investment term structure and market volatility, we find that the correlation between average investment term structure and large price fluctuations is relatively obvious, while the consistency of average investment term structure enhances, large price fluctuations shows a rising trend. But the correlation between average investment term structure and small price fluctuations on short-time scale is not evident.