股票收益率的混合Gumbel分布研究

The mixed Gumbel distribution of stock returns

随机理论认为股票收益率服从正态分布,但大量研究表明,股票收益率等金融时间序列具有"尖峰厚尾"反正态性.建立在极端事件风险理论的基础上,提出了由原始分布和尾分布组成的混合分布模型.最后对深证成分A指的日收益率进行了混合分布拟合,结果显示混合分布不仅比正态分布在效率上有了明显的提高,而且能较准确地度量股票的风险.

The returns of stock market should obey the normal distribution in light of the stochastic theory. However, the financial time series, such as stock index returns, are of high peaks and heavy tails. On the basis of the extreme events theory, a mixed distribution model is presented. The mixed distribution is fitted to stock index returns from Shenzhen stock exchange. It＇s demonstrated that the mixed distribution shows more efficient than the normal distribution and can accurately measure stock＇s risk.