近年来伴随着金融市场广度与深度的不断拓展,频发的金融风险对世界经济及金融市场造成了巨大损失(如美国次贷危机),学者和投资者越来越关注规避小概率巨额风险的最优投资决策及有潜力风险资产遴选方法的研究.文章就此开展了如下研究:首先以损失超过VaR部分的条件期望CVaR作为投资者愿意承担风险的上限,改进投资预算约束为非紧约束,提出了基于巨额损失波动性的投资组合模型.数值试验验证了模型具有良好的收敛性,即使在生成较少数量的情景下也能快速收敛;当投资者对最低期望收益率要求不高时,不必全额投入预算资金就能满足投资者对预期收益的要求;随着投资者对最低期望收益率要求的提高,更多预算资金被投入可能带来更高收益的风险资产,资金预算约束逐渐趋于紧约束;模型给出的最优投资决策在样本外各滚动窗口测试中均实现了较高收益,但发生巨额损失的波动程度却显著降低,达到了控制小概率极端风险的目的.其次,结合常规基本面分析法和聚类分析技术,提出了风险资产的遴选方法.该方法适用于跨市场跨行业不同品种间风险资产的筛选,可兼顾同一类别内资产的同质性及不同类别资产间的异质性,以此达到分散化解风险的目的.实证研究表明,该方法遴选出的“少量”风险资产在各项评价指标上具有明显的优势,聚类技术的引入大大降低了投资者选择资产的难度. In recent years, with the continuous expansion of the breadth and depth of financial markets, the frequent financial risks have caused huge losses to the world’s economy and financial markets (such as the U.S. subprime mortgage crisis). Scholars and investors are increasingly concerned about evading the huge risks of small probabilities Optimal investment decision-making and potential risk assets selection method.The article studies the following aspects: firstly, CVaR is expected as the upper limit of investors willingness to bear the risk and the investment budget constraint is non-compact with the condition of loss exceeding VaR A portfolio model based on large loss volatility is established.Numerical experiments show that the model has good convergence and can converge rapidly even when a small number of scenarios are generated.When the investors do not have a high demand for the minimum expected rate of return, Invest in the budget funds in full to meet the expected return on investors’ requirements. With the increase in the minimum expected return rate of investors, more budget funds are put into risky assets that may bring higher returns, and the capital budget constraints gradually In the tight constraint, the optimal investment decision given by the model achieves a higher yield in each rolling window test outside the sample , But the volatility of the huge loss is significantly reduced, reaching the purpose of controlling the extreme risk of small probability.Secondly, combined with the conventional fundamental analysis and clustering analysis techniques, the paper proposes a method of risk asset selection.The method is suitable for cross-market The screening of risk assets across different industries can take into account the homogeneity of assets in the same category and the heterogeneity among different types of assets so as to achieve the goal of decentralizing and resolving risks.The empirical studies show that the “ A small amount of ”risky assets have obvious advantages in various evaluation indexes. The introduction of clustering technology has greatly reduced the difficulty of investors in choosing assets.