Research On Simulation Experiment Of Continuous Double Auction Financial Market Based On Multi-Agent Model

Financial market is a kind of complex adaptive system that initiated by the construction and emergence of a numerous investors. Since the complexity and adaptability of the financial market, the traditional modeling approach, e.g., mathematical modeling and statistical modeling have been unable to meet the requirements of the current financial research. With the development of the multi-Agent based computer simulation technology, it is possible to build an artificial financial market to study the financial problems. So Agents can be used to imitate the behavior of human investors, with that people can simulate the markets in the microcosmic perspective or the way from bottom to up, through the imitation interaction through a variety of Agents.Most existing artificial financial markets rarely consider investors' bounded rationality in actual markets, but all based on the assumptions that the investors are all fully rational and pursue the utility maximization. In the view of these weaknesses, this paper proposes a new artificial financial market model based on continuous double auction. In this model, Agents are classified into two types, fundamentalists and chartist, and several psychological effects that described in behavioral finance, e.g., anchoring effect, mental accounting and disposition effect, were also introduced into the decision process of Agents to perform their bounded rationality.Further, on the basis of the mathematical model of this financial market, this paper have implemented its simulation system, using the java object-oriented programming language and Anylogic simulation software, which have multi-Agent modeling tools and flexible animation framework, to implement this artificial financial market simulation system.Finally, according to some designed experiments on this simulation system, we found several basic features, e.g., fat tail and nonlinear of the return series of this market model, which also exist in actual financial market. The experiment results show that our model has a certain degree of rationality and can be used to study the financial markets. Moreover, this paper designed two experiments to study the relationship between the tick size and the trading volume and which types of Agents will occupy an advantage in the market, respectively. The results are as follow: smaller tick size will rebound to reduce the bid-ask spread, and on the contrary, with the enlargement of tick size, the daily trading volume will decline significantly; the fundamentalist would be the majority and can acquire more benefits tan chartist in an enclosed market.