均值-CVaR模型的稳健构建方法及其比较研究

The Construction and Comparison of Robust Mean-CVaR Model

传统均值-CVaR模型具有不稳健性,当样本数据存在离群值时,传统均值-CVaR模型获得的投资效果可能会偏离实际情况。针对这一现象,文中将稳健统计思想与传统均值-CVaR模型相结合,构建出稳健均值-CVaR模型以削弱或消除离群值的影响,从而获得较为可靠的投资效果。为了说明稳健改进方法的有效性和可行性,文中进行了模拟实验和实证分析,结果均表明:当数据中不存在离群值时,传统和稳健均值-CVaR模型获得的投资效果基本保持一致;但当数据中存在离群值时,传统均值-CVaR模型获得的投资效果与实际不符,而稳健均值-CVaR模型获得的投资效果仍保持着良好的一致性,说明稳健均值-CVaR模型对离群值具有较好的抗差性和抗干扰性。

Traditional Mean-CVaR model isn’t robust,which means when there are outliers in the data,the investment effect of traditional Mean-CVaR model may deviate with reality.Aiming at this situation,this article combines robust statistic with traditional Mean-CVaR model to construct robust Mean-CVaR model and to reduce or eliminate the influence of outliers and to obtain more reliable investment results.In order to illuminate the effectiveness and feasibility of robust method,simulation experiment and empirical analysis are carried out.The results indicate:When there are no outliers in the data,the investment effect of traditional and robust Mean-CVaR model are basically consistent;when there are outliers in the data,the investment effect of traditional Mean-CVaR model isn’t consistent with the reality,while the investment effect of robust Mean-CVaR model still remains a good consistency with reality,which means robust Mean-CVaR model has a good robustness and anti-interference towards outliers.