中文摘要：

Copula函数是处理变量间相关结构的有力工具。特别在实际的金融领域中,变量间的相关结构往往比较复杂。如果用单个Copula函数来处理,具有一定的局限性。混合Copula（M-Copula）函数是把不同的Copula函数混合在一起,这样能够更好地描述变量间的相关结构。文章主要运用由Gumbel、Clayton和Frank组成的混合Copula模型对上证A指和成份A指的相依结构进行建模,采用非参数核密度方法估计边缘分布,然后用EM算法求出混合Copula函数的权重及相关参数,实证结果表明：混合Copula模型更能够准确地描述两个市场之间的相依结构,且两个市场的上尾相依关系要强于下尾的相依关系。

英文摘要：

Copula function is a powerful tool for handling the correlation structure between variables.Especially in the actual financial field,the correlation structure between variables is often more complicated.If you use a single Copula function to deal with the correlation,there are certain limitations.The mixed Copula function is mixed with different Copula functions,so that it can better describe the correlation structure between variables than the single Copula.The mixed copula model made of Gumbel、Clayton and Frank is applied to establish a model of dependence structure between Shanghai A-Shares Index and Syhthetical A-Shares Index in the paper.Firstly,Non-parametric kernel density method is used to estimate the distribution function of Copula,then in order to calculate the weight and the related parameters using EM algorithm.The results indicate that the mixed copula model can describe the dependence structure between two markets more accurately,and the dependence relation of upper tail of two markets is stronger than the dependence relation of lower tail.