Designing Trade Mechanism For Stock Market Based On Agent-based Computational Finance

Traditional asset pricing theory deem trading mechanism as exogenous variables that have no relationship with price forming process.As the development of the market microstructure theory, more and more researches broke the frame of traditional CAPM and APT assumptions, such as "Efficient Market" and"investors completely rational".Investor behaviors and Transaction cost and so on also affect the liquidity,volatility and efficiency of stock market.Closing price mechanism as an important part of trading mechanism that have an important impact on the stock market.Closing price assume itself can represent stock price on that day for it is easy to collect, traditional closing price mechanism based on the price of last day,but can be manipulated easily,also has defects in represent investors’expected value.In the long run,this will inevitably lead to stock price deviating from fundamentalist value and investors’expected value.Different financial assets has its unique properties,how to design suitable mechanism for these financial products is a important issues the stock maket regulators must face.The study use the modeling advantage of Computational Experiment Finance,constru-cting the simulation financial market with investors behavior model and trading mechanism model,with controllable financial experimental redesign trading mechanism.This thesis firstly put forward the concept of expect price and instead of original closing price mechanism.Comparative analyse new mechanism’s impact on stock price and market quality.Keeping the quantities of all investors’demands in one trade day unchanged while allowing their arrival orders randomly altered, we investigate the statistical properties of price series under two trading mechanism-continuous double-auction and only one call auction executed in one trade day, and then redesign a closing price mechanism with satisfying properties. Theoretical analysis and financial experiments find that:1) by altering the arrival orders of risk demands, both the stock closing price and the intraday price behavior are changed radically, which further effects the long-term price trends;2) if the conventional closing price is replaced by the final transaction price under only one call auction executed in one trade day, new closing price is closer to investors’expectations about assets fundamentals value than the old one; though the volume in one trade day under new closing price mechanisms is below to that under the traditional one, new mechanism is helpful for reducing the systematic risk and manipulation risk of stock market; besides, some anomalies in statistical properties of return distribution under new mechanism, such as fat tails, excess kurtosis and volatility clustering, are not so obvious as those of real stocks. These findings suggests that:a reform about the closing price mechanism should be conducted, i.e., the conventional closing price which uses the final transaction price of a trade day should be replaced by the transaction price under only one call auction executed in one trade day.