Entropy based robust portfolio

Abstract Whether entropy is more suitable to measure risk of portfolio or the portfolio diversification, actually, is an endless controversy. So, as the risk measurement and the portfolio diversification measure, entropy is respectively introduced to MV model, obtaining entropy based portfolio models. Meanwhile, higher moments (skewness and kurtosis) are recommended to relax the assumption of normal distribution and reflect the extreme events. Furthermore, consideration of robust optimization approach estimates the uncertain input parameters in these models; subsequently, entropy based robust portfolio models with higher moments are constructed. Moreover, multiobjective particle swarm optimization is applied to tackle these sophisticated portfolio models. Eventually, empirical comparisons indicate that entropy is more suitable to diversify the portfolio; importantly, robust portfolio models taking entropy as the measure of the portfolio diversification can provide the optimal portfolios, and significantly improve portfolio performances. Additionally, higher moments should not be ignored in the entropy based portfolio models.