Portfolio And The Particle Swarm Algorithm For The Method

With the rapid development of economy, our country's financial market and financial system are improved continuously. As a major public investment, securities are more and more popular for people with the growing of the national average per capita disposable income in recent years. Every investor hopes to select good securities and to pursuit high yields with low-risk. There are many models and methods can help investors to achieve the expectation in fact. This dissertation takes advantage of the factor analysis to choose and evaluate the securities, and proposes the optimal combination investment model for the assembly. After that, uses particle swarm optimization (PSO) to solve the proposed model. The work in the paper has a definite practical significance for the stability and development of China's financial market, and supplies investors with valuable decision-making basis. The main work of the paper can be summarized as the following:Firstly, the dissertation explains the significance of selecting and evaluating securities, introduces the factor analysis, and uses it to select and evaluate the stocks of listed company.Secondly, the dissertation analyses the shortages of the Markowitz composite model, introduces the kurtosis factors and proposes the Mean-variance-kurtosis Model. Based on the analysis in theory, we can obtain a result that the optimal-improved model is closer to the reality, and can supply better instructional basis for investors.Thirdly,the dissertation introduces PSO to solve the Mean-variance-kurtosis Model, and selects the stocks of Shenzhen bourse to make a simulation and comparison analysis.Finally, the dissertation proposes the deficiency and the work need to be done in future.There are two innovation points here: (1) the dissertation leads kurtosis factors into Markowitz's portfolio model, which makes up for the shortage of the original model, and makes the model closer to the actual investment; (2) the paper uses PSO to solve the Mean-variance-kurtosis Model. This method is simple and easy to achieve, and can not get in the local optimal easily for its random searching mechnism.