Mean-Variance-Approximate Skewness Portfolio Slection Model And Empirical Analysis

As the foundation of the modern finance theory and investment theory, port-folio selection theory studies how to allocate capital to different assets in order to diversify the portfolio's system risk while maximizing the return. In 1952, Markowitz introduced the famous Mean-Variance (MV) model based on measur-ing the portfolio risk by the variance, which provided the basis for the modern investment theory and opened an era for the quantitative financial analysis.The classical Markowitz's Mean-Variance model in modern investment sci-ence employs variance as risk measure to formulate the volatility of return, while it ignores the asymmetry of the return distribution. The introduction of skew-ness as asymmetry measure, however, results in a nonconvex three-order poly-nomial programming problem which is difficult to solve. In this thesis, we pro-pose a Mean-Variance-Approximate Skewness (MVAS) model which measures the asymmetry by the ratio of positive semi-variance to negative semi-variance. This model effectively reduces model complexity and improves skewness of the invest-ment portfolio. We study the empirical performance of MVAS model based on Chinese stock market, and obtain some conclusions. Firstly, the portfolio mod-els which consider the asymmetry of the return distribution produce a portfolio with high skewness, which is more likely to gain excess return. Moreover, the out-of-sample performance of different models under both short and long periods shows that MVAS model outperforms MV and MAD models when the market has non-normal feature.The thesis is organized as follows:In Chapter 1, we introduce the basic the-ory and background of portfolio selection. In Chapter 2, we present several high moment portfolio optimization models including Mean-Variance-Three Moment model, Mean-Absolute-Deviation-Three Moment model and Mean-Variance-Four Moment model. In Chapter 3, we study Mean-Variance-Approximate Skewness (MVAS) model which measures the asymmetry by the ratio of positive semi-variance, to negative semi-variance. In Chapter 4, we compare the performance of MV model and MVAS model. Extensive empirical experiment is carried out using real data from Shanghai Stock Market. Finally, we conclude the thesis in Chapter 5 with some concluding remarks. We summarize the main results of the thesis and suggest some future research directions.