建立在均值—方差分析框架下的组合投资决策,需要较强的正态分布假设,难以准确刻画与分散非对称与极端尾部风险。为此,文章考虑均值-VaR模型,将模型求解过程转化为一个分位数回归问题,给出了均值-VaR模型求解新算法。使用沪深300指数中的60只成分股进行了实证研究,验证了算法的有效性,并将基于分位数回归的均值-VaR模型与均值—方差模型进行了对比,发现前者能够很好地分散尾部风险。
The traditional mean -variance portfolio selection model need a rigorous normal distribution as- sumption. It is difficult to accurately describe and diversify the asymmetric and extreme tail risk of financial assets. So far, we consider the mean - VaR model and propose a new algorithm for its solution tile regression approach. For illustration, formance of the mean- VaR model based model. through quan- we use 60 stocks in Shanghai and Shenzhen 300 index. The per- on quantile regression is superior to the traditional mean -variance