Research Of Tail Dependence Of Financial Data

Along with the finance research area developing, the relevant analysis becomes more and more important in the financial application. Investment analysis, assets' diversification and pricing and financial risk measure all involves to relevant analysis. Since 90's the frequent world scope financial crisis, relevance research was pulled to guards against the disaster and showed the great significance. But the former research usually all concentrates in to the degree of relativity, and neglected to related structure or the tail dependence in the money market. In fact, the same relativity of two pair random variables possibly can have the different related pattern and the special part characteristic. Therefore using the degree of relativity or the linear correlation describe dependence of correlation during the random variable is not comprehensive.Using the copula function technology, may take the relativity and the correlation pattern organically into together. As the connection of marginal distribution function, the Copula function may not only reflect the relativity of random variable, but also may describe the random variable's related pattern well. Therefore we may use the different Copula function to describe the different related pattern. The predecessor already provided and applied the copula function in the financial domain and has summarized a general frame. And has carried on the thorough analysis and the application to several common copula functions. This article's emphasis studies the random variable using the copula function to describe tail dependence. This article has first carried on the detailed introduction to the copula function, then has a induction to the tail dependence, and uses the copula function to express it. At the same time chi chart is a traditional method, which was applied in the correlation judgment. Finally carries on modeling processing based on the judgment result to the tail dependence. To the analysis method, this article continues to use the predecessor's copula function method frame, but this article emphasizes the flexibility and adaptability of different dataduring modeling. Therefore this article fully used the convex combinatorial property of copula function to construct a flexible copula function combination, and combine the different copula function's predominance together. Finally we also have proven this article's flexible method is better.This article's conclusion is that flexible method can be a good simulation of real data relevant situation, especially tail dependence. When we compare with the normal copula function method and the t-copula function method in predecessor literature, our model show more predominance than others.