摘要
针对传统Pearson线性相关系数与Granger因果分析法的不足,采用一种特殊的相关性分析方法—Copula函数方法对沪深股市相关结构与相关模式进行研究。首先用核密度估计方法对Copula函数的边缘分布进行估计,再结合秩相关系数对数据拟合较好的Copula函数进行选择,最后用离散L2范数评价方法对其拟合程度进行检验。研究发现,t-Copula可以较好地拟合沪深股市的日收益率序列,沪深股市日收益率序列呈现出较强的相关性以及对称的尾部相关性,当沪深两市出现大幅震荡时,两市收益率的协同作用将大幅增强。
For the shortage of traditional Pearson linear correlation coefficient and Granger method of Cause and effect analysis,a special correlation analysis method — the method of Copula function was used to study the dependence structure and correlation pattern of Shanghai and Shenzhen stock market. First,the method of kernel density estimation was used to estimate marginal distributions of Copula function,then the rank correlation coefficient was combined with to choose a better copula function that can fit the data well,at last discrete L2 norm was used to fit degree test. The study found that t-Copula can fit the Shanghai and Shenzhen stock market's daily return rate sequence better. The return rate sequence shows a strong correlation and symmetrical tail dependence between Shanghai and Shenzhen stock market. When the Shanghai or Shenzhen stock appear sharp situation,the synergistic effect will be greatly enhanced.
出处
《四川理工学院学报(自然科学版)》
CAS
2016年第2期70-74,共5页
Journal of Sichuan University of Science & Engineering(Natural Science Edition)
基金
陕西省教育厅专项科研基金(14JK1299)