Portfolio Optimization Model Based on Conditional Value-at-Risk

Based on an algorithm proposed by R T Rockafeller and S Uryasev, a portfolio optimization model, mean-conditional value-at-risk model, was set up. This model measures riskwith conditional value- at-risk (CVaR) instead of standard deviation. It was optimized by using Matlab software and choosing six stocks in Shanghai and Shenzhen stockmarkets in China as a portfolio, and efficient frontier and investment proportion of the portfolio were obtained. By comparing them with the ones obtained using the traditional mean-variance (MV) model, the result indicates that the efficient frontiers gained by the two models are almost identical and also close to overseas research results, but there is a difference between the optimal investment proportions based on the mean-CVaR model and the MVmodel.