Portfolio selection based on relative entropy coherent risk measure.

This article optimizes stocks portfolio through a new proposed coherent risk measure called relative-entropic risk measure. And then the ability of the optimization model is tested, which is the test of the discrimination for risk of the risk measure. Firstly, it introduces the consistence with stochastic dominances for three kinds of quantile-based risk measures: VaR, ES and relative-entropic risk measure. It is pointed that ES only utilizes local information as VaR, and is consistent with stochastic dominances lower than second-order. However, relative-entropic risk measure utilizes the whole information to measure the risk, and is consistent with stochastic dominances of higher orders. So, it is most powerful for discrimination of risk. Then, Spearman rank test is used to test the discrimination for risk of the risk measure. Lastly, four kinds of portfolio optimization models based on standard error, VaR, ES and relative-entropic risk measure are used in Chinese markets. The results turn out that all the performance indexes of portfolios optimized by relative-entropic risk measure are the best, which reveals that it has the highest discrimination of risk in the three risk measures. [ABSTRACT FROM AUTHOR]