改进Copula对数据拟合的方法

A Method of Improving Copula Fited to Data

给出相关结构Copula、秩相关系数Spearmanρ与Kendallτ和尾部相关系数η,以及这三个关联性度量与Copula之间的关系,各个相关系数的估计方法.在一个Copula族内进行适当变换,得到新的Copula,使得能更好地拟合样本的各个相关系数.最后,以沪、深日收盘综合指数为例,讨论了二个股市波动率的相关性,建立了一个较好的数学模型.

The conceptions of dependence structure, copula, coefficients of rank dependence including Spearman ρ and Kendall τ and coefficient of tail dependence η are presented. The relations between copula and three dependence coefficients and algorithm for calculating the estimations of these coefficients also are shown in this paper. Moreover, we introduce a transformation of copula and permit to fit the dependence coefficients in a better way. In the last, as examples concentrating on studying dependence of fluctuation associated to the data of intra-daily close index on the Shanghai and Shenzhen stock market are given.