The Fitting For Fund Portfolio Density Function With Mixed Pair Copula-Kernel Estimation

In recent years.functions of Pair Copula are widely used in financial field.It is used to analyze the nonlinear relationship among these assets. This paper construct-ed a mixed Pair Copula-kernel estimation model,it is used to fit a combined prob-ability density distribution function for the series of fund portfolio’s yield rate.The model uses method which is kernel density estimation to estimate the marginal distri-bution.Compared with GARCH,it doesn’t need the assumption for the distribution of residual sequence.So it can reflect more fluctuation characteristics of time series.We choose the stock yield’s series who are among the top five in Guotai small caps as the research object in empirical study,use both kernel estimation and GARCH model to estimate the marginal distribution,and compare the result with the two methods.The conclusion that kernel estimation is better than GARCH model is proved.In addition,it gains Pair Copula function according to Vines analyze on the basis of Copula func-tion.The model chooses the stock yield’s series who has stronger correlation with other stock yield’s series than others as the center element. The model constructs Mixed Pair Copula function,it chooses different Copula function when it chooses two dimension Copula in every tree,uses the best two dimension Copula to fit from the data itself.On the basis of marginal distribution series,it is used to estimate the combined probabili-ty density distribution function of the fund portfolio through both mixed Pair Copula function and non hybrid Pair Copula function.Then.compare with the results,we get that mixed Pair Copula function is better than non hybrid Pair Copula function.