A Comparative Analysis of Stock Portfolio Constructed on Computational Intelligence and Time Series Model with ISCI

• Main Authors: LIU,KO-FANG, 劉可方
• Other Authors: CHANG, JUI-FANG
• Format: Others
• Language: zh-TW
• Published: 2016
• Online Access: http://ndltd.ncl.edu.tw/handle/03569378979657369185

碩士 === 國立高雄應用科技大學 === 國際企業研究所 === 104 === In recent years, designing stock investment portfolio has become a popular subject. Investors are interested to know how much reward and profit can the portfolio brings to them. Therefore, this paper focuses on studying the portfolio construction. Our experiment contains two stages: the first stage is the stock selection; and the second stage is the stock investment portfolio construction. In the first stage, the Investment Satisfied Capability Index (ISCI) is utilized to select stocks and measure every individual stock performance. In the second stage, the time series models and the Interactive Artificial Bee Colony (IABC) , Particle Swarm Optimization(PSO) algorithm are applied to construct the stock investment portfolio in order to receive the maximum reward. The experiment data includes five-year daily rate of individual stock return from 2011 to 2015. Our experimental results indicate that planning the investment weight distribution via IABC, PSO gains more stable rewards than via time-series models such as GARCH and Stock market.