衡量金融风险相关性的Copula熵方法

正态检验的不理想和偏度及峰度的存在使得用基于多元正态分布的假设及线性相关系数研究相关性受到质疑,为此应用信息熵结合Copula理论建立了Copula熵函数.与信息论中互信息等相关性衡量指标相比较,其具有不受维数限制、有量纲、能捕捉非线性相关关系等优点.结合经济圈理论进行了数据验证,表明了该方法的可行性和有效性.

Copula entropy method for measuring correlation in finance risk

In response to the existence of the denial of the normal test and skewness as well as kurtosis, the research on correlation based on the assumption of multi-dimensional normal distribution and linear-dependent coefficient may not be accurate and proper. Therefore, the Copula entropy function is founded by combining Copula theory with the entropy of information. Compared with the mutual information in information theory and the correlation coefficient, the Copula entropy function has the advantages that it is not confined to dimensions and has dimension and the ability of taking non-linear correlation relation. Then, the experimental data is validated by combining with the economic circle theory, which shows feasibility and validity.