Firstly,we compare the mean-variance model using variance to measure risk,the mean-absolute-deviation model using the absolute deviation to measure the risk,and the mean-semi-variance model using the semi-variance to measure the risk.The comparison shows that with the same return,the semi variance as the risk measurement is most consistent with the psychology of investors.Then we employ the entropy to construct the mean-semi-variance-entropy model.The analysis of a numerical example shows that our proposed model is more comprehensive for it is concerned with all kinds of uncertain factors in the stock market,and can provide more secure investment plans for the investors.

%U https://zkxb.jsu.edu.cn/EN/10.3969/j.cnki.jdxb.2017.03.017