Research On Stylized Facts In Continuous Double Auction Trading Mechanism

A great number of empirical studies on the stock market have revealed that several stylized facts is common in stock market of different time span and different countries,including fat-tailed return distribution and volatility clustering.Financial theories based on investor rationality and efficient market hypothesis cannot provide any solid explanation of such phenomenon.The development of Agent-based Computational Finance provides a new method to explain these stylized facts.Agent-based Computational Finance defines investors as adaptable agent,using IT to simulate market microstructure and investor behavior,and then make the artificial financial market.Explain the fluctuation of market price and its reason at macro level by conducting experiments in artificial financial market.Scholars at home and abroad have made some achievements in related research,but there are some limitations now.For example,most models are too complex,and the interpretability and analyticity are deficient,few study take account of both market microstructure and investor behavior,and no research has been done to analyze the formative mechanism of stylized facts.This paper based on the perspective of Agent-based Computational Finance,we studied the influence of the Continuous double auction(CDA)mechanism and feedback trading to fat-tail and volatility clustering.Firstly,we establishes a simple artificial stock market model with zero-intelligence traders,and then expand on it,for the expansion of artificial stock market model has a certain amount of feedback traders,the other with certain proportion of feedback traders.By comparing the kurtosis of yield series distribution generated by these three stock market models,we systematically analyze the influence of market microstructure and feedback trading behavior on heavy-tailed return.Then,based on the GARCH model,we compare the GARCH fitting parameters produced by three artificial stock market models,and systematically analyze the influence of market microstructure and feedback trading behavior on volatility clustering.Our research mainly draws the following conclusions:(1)The CDA mechanism itself can produce heavy-tailed return phenomenon,and it is related to the density of the instruction arrival.With the increase of the density of the instruction arrival,the peak fat tail degree of the yield distribution gradually weakened until it reached the normal distribution.(2)The CDA mechanism can generate the phenomenon of volatility aggregation itself,but it is relatively weak.The degree of volatility aggregation is not significantly related to the instruction arrival density.(3)When there is a certain proportion of conditional heteroscedasticity trader,with orders to the increase of density,peak fat tail phenomenon no longer disappear gradually,but with the absolute number of conditional heteroscedasticity traders increased peak fat tail phenomenon also strengthened,leptokurtosis phenomenon visible in this period was mainly caused by the feedback trading behavior.(4)The behavior of feedback transactions can aggravate the phenomenon of volatility aggregation.Moreover,the volatility aggregation of returns series is related to the number of feedback traders.With the increase of the number of traders with feedback trading behavior,the volatility aggregation of returns is enhanced.