The Minimax Model For Portfolio Selection And Truncated Aggregate Homotopy Algorithm

In the stock market,to achieve more profit and disperse the risk,many investors always invest his found in some different securities,the so called portfolio.How to allocate found in these selected securities,this needs some proper models to determine,and this is a core problem in portfolio.Unremitting efforts have been making since Markowitz constructed the mean-variance model in 50's.Because of the history and system,the research on advanced portfolio in our country started rather late.But after more than ten years endeavor,the scholars in our country have made much advancement in the area.In view of how to determine the investment proportion,the mean-variance model is based on an assumption that the return on assets obeys normal distribution.In order to overcome this limitation,this paper adopts the minimax model for portfolio selection based on the sample data,and uses absolute deviation to measure invest risk.It can avoid the minimal return and the maximal risk for the conservative investor.Then this paper extends the truncated aggregate homotopy algorithm for solving the unsmooth optimization problems,and verifies the investment proportion based on closing price.The results show that the investment proportion is obtained more quickly by the truncated aggregate homotopy algorithm,and the investment decision makes a stable profit.