基于Copula函数和非参数估计的尾部相关性

Tail-dependence Correlation Based on Copula Function and Non-parameter Estimation

给出基于Copula函数的尾部相关性的定义和性质,采用非参数方法估计尾部相关系数.结合数据得出上证指数和深圳指数的尾部相关系数和对应图形比较,可知两种股票的上尾比下尾相关性强.此相关系数反映了上证指数与深圳指数在极端值处同时小于或同时大于某个数值的概率大小.

The paper gives definition and character of tail dependence based on Copula function, uses the non-parameter estimation to estimate tail-dependence coefficient. Combine the data of compare th lower tail-d SZZS and SZZZ, we obtain tail-dependence coefficient and corresponding graphs, em and then conclude that the upper tail-dependence correlation is stronger than ependence correlation. The dependence coefficients reflect probability when extreme value is more than or less than some numerical value at the same time.