Integrating Principle Component Analysis with PSO for Portfolio Selection

• Main Authors: Ju-Ping Hsu, 許茹蘋
• Other Authors: 邱垂昱
• Format: Others
• Language: zh-TW
• Published: 2009
• Online Access: http://ndltd.ncl.edu.tw/handle/zyw538

碩士 === 國立臺北科技大學 === 工業工程與管理研究所 === 97 === In the first phase of the use of principal component analysis into the traditional financial indicators, in the traditional financial indicators, financial indicators to select multiple variables, using principal component analysis to simplify a number of related variables into a few principal components is not related, to a small number of principal components to represent the original number of variables. Principal component will be a small number of variables were calculated before the high point of 10 stocks of the next quarter as the subject of the investment portfolio. The second phase by the use of principal component analysis of 10 selected investment subject to particle swarm optimization algorithm to carry out asset allocation, taking calculated risks, returns the best fitness value (Max Sp). Finally, the results obtained, with the weighted index and the Taiwan 50 index for comparison. In the empirical study, principal component analysis in stock selection, combined with particle swarm algorithm to adapt to randomly strike the greatest value of the results obtained, the results from the asset allocation between 2006 and 2008 are maintained at a positive rate of return, rate of change than the weighted index and the Taiwan 50 Index smooth; Sharpe value, the change than the weighted index and the Taiwan 50 index, the average values are being Sharpe; risk, which is lower than the weighted index and the Taiwan 50　Index .　Consolidated results available, the investment target in the scope of acceptable risk investors, the more will be paid over.