The Financial Complexity:Study On Price Behavior And Market Mechanism On Model

With the evolution of economy structure, the virtual economy becomes more and more infusive. The three big finance crises in1987,1998and2008are of great influence for the world. However, it’s hard to explain the stylized facts in the finance market for the classic finance theory. So taking the fact that finance market is a typical complex system into account, this paper explore the auto-dependence of the volatility and build two agent-models to investigate the formation mechanism of the stylized facts and the loss aversion based on the complexity theory. The main research results and innovation of this paper are itemized as below.1. The controversy exists in two cases of the study of the market volatility. First, there is no evidence for auto-correlation between the successive returns, but obvious evidence exists. Second, whether the GARCH model could depict the long memory property is also controversial. So this paper studies the auto-dependence of the volatility. It’s found that, there is a linear dependence between the logarithms of successive average volatility with a small observation scale. With the increase of observation scale, the linear dependence becomes weaker, while the nonlinear becomes stronger.2. Similar cased is also found in GARCH process. But after a quantitative study, the deference is found between real data and GARCH model. What’s more, the volatilities of different periods have distinct influence on the correlation structure, and GARCH model undervalues the long memory property.3. CB model bridge the fat tail and herding effect successively, but it doesn’t model the detail process how people interact each other. Considering this situation, under the connection of the interaction process of people, the nonlinear influence of institute investors and the evolution of stock network, this paper builds an agent model to describe the formation of herd behavior and the evolution of stock market. The model is simulated under different initial conditions and parameters. All the results reproduce the volatility clustering and peak-fat-tail. The statistic value accords to the empirical:the power-law exponents between the peak value of return probability distribution and the time scales ranges from0.579to0.747, and that exponents between the accumulation distribution and the return on the tail is close to3. Besides, the extent of volatility clustering in our produced price series is close to that of S&P500and locates between NASDAQ and HSI.4. This paper builds an agent-model to investigate the common formation mechanism of irrational behaviors and stylized facts, which is mentioned in little literature. The model reproduces the formation of the loss aversion. The detailed analysis suggests that significant law of diminishing marginal utility is very important for the formation of loss aversion. And the behavior that people may follow the successful persons is also of help for the loss aversion.