Mean-variance Portfolio Model And Empirical Research Based On Yield Forecast Of Multi-factor

Since the late80’s and at the beginning of the90’s in the last century, the securities market has begun to take shape slowly in China. The whole scale of Chinese securities market improves gradually and it forms a more perfect system. However, the large-scale financial market is extremely complicated. Each investment activity has a high risk and the yields on assets have a high level of uncertainty. Majority of investors don’t have a correct understanding of the risk of the investment, blindly into the market, finally led to the failure of investment. The most important reason of failure can be attributed to be lack of scientific investment strategy and weak risk awareness investors themselves. It becomes the core problem that how to invest funds for a reasonable asset allocation as to investors, when the stock price is fluctuate, so investors can not suffer great losses, or to guarantee a certain profit.Based on these, the establishment of the invest portfolio optimization model of this article is based on the Markowitz mean-variance model, and is combined with the multi-factors regression model proposed by Goldfarb and Iyengar. And the article selects a closed-end fund to conduct an empirical analysis which performance is well. Firstly, the article has an overview of the classic portfolio theory, principal component analysis theory, multiple linear regression model and theory. Secondly, the six selected market indexes are integrated into the first factor and the second factor by the use of principal component analysis method. The two factors are applied into the multi-factor model proposed by Goldfarb and Iyengar to get the estimated values of parameters μ and V, which are expected yield value and Influence coefficient value. At the same time, residual variance D is get though the use of non-biased estimated method. After, the paper analysis the date of "Da Cheng" closed-end funds from2008-2011and compares the actual weights and the optimized weights.The result is that optimized yield than the actual yields is higher. Finally, the paper summarizes and points out the inadequacies and further research directions.