基于均值-CVaR的投资组合优化问题研究

【作者】 徐雅娉；

【导师】 刘喜华；

【作者基本信息】 青岛大学 ， 金融学， 2010， 硕士

【摘要】 投资组合优化研究的是投资者如何通过合理的资金分配来达到既定收益下风险最小化或者既定风险水平下收益最大化,即如何选择最优资产组合。目前,该理论作为主要的分析工具已经被广泛应用于投资和理财中。投资组合优化理论不仅可以指导投资者做出科学的投资决策,而且还可以使金融投资实行真正的科学化管理。本文在已有研究成果的基础上,基于CVaR方法,建立均值—CVaR投资组合优化模型,然后,从上证50指数样本股中选取了10只股票构成一个投资组合,分析了在同样置信水平下均值—方差模型和均值—CVaR的表现,研究了加入无风险资产之后对均值—方差模型和均值—CVaR模型有效前沿的影响。研究结果表明,CVaR能够更好地度量风险,尤其是当资产收益率不满足正态分布时；随着置信水平的提高,CVaR值不断变大,说明CVaR方法对尾部损失测量的精度更高；无风险资产的加入会使得均值—CVaR模型有效前沿发生变化。从整体上看,基于CVaR的均值—CVaR优化模型,无论是从精度上还是广度上,在对投资组合的风险度量和风险控制方面都比均值—方差模型或是均值—VaR模型具有更好的适应性。更多还原

【Abstract】 Portfolio optimization studies the investors how to achieve minimum risk under the defined benefit level or maximum benefit under the given level of risk through reasonable distribution of funds. That is how to select the optimal portfolio. At present, the theory as the main analytical tool has been widely used in the investment and financial management. Portfolio optimization theory can not only guide investors to make scientific investment decisions, but also the financial investment is a real scientific management.In the basis of existing research results, the paper based on CVaR ways to build mean-CVaR portfolio optimization model. Then select 10 stocks from the sample stocks in the SSE 50 Index to research and analyze the results. Then, we compared the mean-CVaR model with the Mean-Variance model, choose a better measure under the same confidence level. The results show that CVaR can better measure risk, especially when the return on assets does not meet the normal distribution;with the increase of confidence level CVaR also increases, indicating that CVaR can better measure the tail loss; risk-free assets would make the mean-CVaR efficient frontier model change. Overall, mean-CVaR is more adaptive than mean- variance or mean-VaR both from the accuracy or breadth in the portfolio’s risk measurement and risk control.更多还原