Study On The Improved Modelof Artificial Stock Market Based On Agent And Scale-Free Network

With the research level of Complexity Science higher and higher, more and more researchers begin to bring the concept of complexity into this field. As a complex giant system, the financial markets have unique complex features, such as peak and fat-tailed, fractal and volatility Clustering. These founds of complex features shook the basis of efficient-market hypothesis (EMH), and have challenged the paradigm of traditional finance research. To discover the reasons behind these features and develop new theories and models, a crossed subject combined Computer Science, Complexity Science and Behavioral Finance appeared. It is called Agent based Computational Economics (ACE). It follows the Bottom-Up method, and takes advantage of capacity of high-speed processing of computer. By this way, it builds Artificial Stock Market(ASM) and achieves a high degree of simulation of the true financial markets, and becomes the"lab of economics"to help the regulation of financial markets and governments'policy making. In addition, the research on the individual decision-making mechanism and inter-group transmission mechanism of influence shows that, not only the asset price finding and forming mechanism can be revealed, but also the market's structure and operation mechanism can be improved. So, the research can help strengthen the market supervision system, standardize the traders'behavior, and protect the interests of investors.First,the paper summarizes the current research on agent based artificial stock market and makes the comment. After that, the related theory is introduced, such as complex science, scale-free network, and the method of agent based modeling. In the process of modeling, the basic issue about the bounded rationality and heterogeneity of investors is discussed, and the rationality of investors in different countries and regions is proved to be different. The basic hypothesis is proposed afterwards. By introducing the theory of scale-free network, the model makes new improvement on the network topology and designs the decision function; it makes the model fits the actual situation better. Combined with the investors'judgment on the stock's basic value and the market's pricing mechanism, the model is simulated by Matlab. The experimental results shows that the scale-free network fits the investors'network of the actual stock market better; some investors working as the"hub"of the market will produced more affect; part of the market's complex features can be explained from the the micro-mechanism design of the market. The research of this paper shows that the Artificial Stock Market (ASM) made by the agent-based financial market method can simulate the true financial market to a certain extent. So, the application value analysis of the model has important pratical significance and considerable relevance.