Abstract: In the traditional mean-variance model, researchers tend to assume that return rate obeys normal distri-bution which is always not satisfied with most questions, and estimate ex pectation value with its average, simultaneously considering the limitations of variance. So, the mean-entropy model based on return weight has been proposed. In this paper, we choose five A-stocks in application. Then, we study the rationality of θ_t present the consistency of the entropy method and variance method, and do some research on comparison between no risk-free securities and risk-free securities. The results show that computing the expectation of return rate based on θ_t the efficient frontier of mean-variance model is still a good parabola, the consistency of entropy method and variance method is proved, and entropy method can better disperse the risk when return rate is equal.

Key words: portfolio selection, mean-entropy model, efficient frontier, return weight, consistency

摘要： 在经典的均值-方差模型中,研究者往往假设收益率服从正态分布,用收益率均值估计其期望。但在实际问题中收益率往往不满足假设,同时考虑到方差度量风险的局限性,从而我们构建均值-熵模型,其中收益率期望用每个时间段收益占总时间段收益权重（θt）来计算。本文通过对深圳A股进行筛选,并从中选取5只股票进行实证分析,然后讨论了θ_t的合理性;熵方法和方差法的一致性;不含无风险证券和含无风险证券均值-熵投资组合模型的比较分析等问题。结果表明：引入θ_t计算收益率期望时,均值-方差模型的有效边界仍是一条较好的抛物线;熵方法和方差法具有一致性;且在相等收益水平下,熵方法能够更好的分散投资风险。

关键词: 投资组合, 均值-熵模型, 有效边界, 收益权重, 一致性