Dynamic Portfolio Optimization Model Research Based On Improved Genetic Algorithm

With the continuous development of China’s market economy and constantly improvement of financial market, the investor’s demands for investment are diversified currently. The main purpose of portfolio optimization is to maximize the return and minimize the risk. Therefore, the research the portfolio optimization problem is of great significance. Recently, the application of intelligent algorithm develops continuously, and the intelligent algorithm is used to solve a portfolio optimization problem which has been agreed by more and more researchers. So, we propose a structured classification based portfolio optimization model and solve the portfolio optimization problem with Genetic Algorithm. A comprehensive analysis of the results is provided, and it is clarified that the structural model is effective on the portfolio optimization problem.At present, most of the domestic and foreign researchers improve the Markowitz Mean-Variance model or carry out certain improvements to the algorithm to solve the portfolio problem, which includes adding the actual constraints, considering transaction costs to the model, and using improved genetic algorithm or other intelligent algorithms. However, there are few researchers taking the classification of securities and the dynamic nature of the stock market into consideration. In this paper, firstly, modern portfolio theory and research are introduced. Secondly, a portfolio optimization model based on structured classification is proposed, and the steps of improved genetic algorithm are given to solve the structural model. Thirdly, the solution method of dynamic portfolio model considering the transaction costs is provided. Finally, according to the historical data of99stocks in actual stock market, the results of simulation experiments finished by the MATLAB data processing software are compared with the mean-variance model. The comparison results prove that the established portfolio optimization model based on the structure of the classification is effective.