The Investigation Of Trading Mechanisms In Continuous Double Auction Stock Market Based On Agent-based Computational Finance

Trading Mechanism is a very important issue in the research areas of financial market microstructure. A suitable trading mechanism can ensure the development of emerg-ing financial market and the financial innovation, especially for the design of new fi-nancial products, which are rich theoretical and realistic value. In view of these points, various design of trading mechanisms in continuous double auction stock market are investigated in this dissertation, which focus on how the trading mechanisms affect the market quality, based on the approach of agent-based computational finance.In this research process, firstly there is a discussion of the methodology and re-search paradigm in agent-based computational finance. Secondly, a continuous double auction artificial stock market computational platform has been built up, which is similar to the real Chinese stock market. And a bounded rationality heterogeneous agent model (BRHAM) has been designed. Another, the agents in this artificial stock market have overreaction and underreaction behavioral characters and they can re-sponse the overnight information impact during the opening time in each day. And the order book is the same as real Chinese stock market which would empty completely after the trading close in each day. Thirdly, a calibration of this artificial stock market has been carried out, in both the traditional stylized facts in the stock markets and other statistical characters. Furthermore, a multi-scale calibration method has been proposed. Simultaneously, using the real stock market's data for parallel calibrating, and all these results of calibration are perfect. Finally, list the benchmarks of market quality and design different 81 experiments with the factor sets of price limit, tick size , tax rate and the disclosure range of order book. And then discuss how the trad-ing mechanisms affect the market quality from the angle of market liquidity, market volatility and market efficiency and their significance, which are based on the me-thods of range analysis and variance analysis.This study found that the artificial stock market computational platform can si-mulate the stylized facts and multi-scale, etc. nonlinear dynamic characters in Chinese stock market. And in the sets of price limit, tick size, tax rate and the disclosure range of order book, the tick size and the disclosure range of order book affect the market liquidity, market volatility and market efficiency more significantly. However, the price limit and the tax rate affect the market quality less significantly. So there should be pay more attention to the tick size and the disclosure range of order book in the scenarios of the simulation correspondingly.