Application of Data Envelopment Analysis and Multi-objective Programming for Optimal Portfolio

• Main Authors: Yang Sheng Chih, 楊勝智
• Other Authors: Wu Tai Hsi
• Format: Others
• Language: zh-TW
• Published: 2010
• Online Access: http://ndltd.ncl.edu.tw/handle/98628464696985353220

碩士 === 國立臺北大學 === 企業管理學系 === 98 === Depositing money in a high inflation economic environment can’t prevent asset value from declining. Thus, how to manage personal investment properly has become a major issue nowadays. In this research, a portfolio optimization model is proposed. Decisions such as selecting stocks, locating capital and adopting trading strategies are to be determined in this model. The mean-variance model was initiated by Markowitz in 1952, which assumes that investors only care about the means and variances of their returns. However, investors usually concern the unexpected declining of stock price much more than that of rising. Thus, we concentrate on the downside risk in our study. This study consists of three stages/decisions. First of all, the fundamental financial analysis and data envelopment analysis (DEA) technique are used for selecting stocks as the candidate of final portfolio. Secondly, three different decision methods are used to determine the distribution of the total capital for investment. These methods include the reference set from DEA, the multi-objective programming technique, and the Mean-Lower Partial Movement model. Finally, the buy-and-hold trading strategy is adopted. Historical data set from January, 2006 to December, 2009 are used to measure the investing performances of the proposed model. The resulting computational results demonstrate that the proposed model (ROI, variance and Sharpe’s index) performs better than the benchmark TSEC Taiwan 50 index and other similarity benchmark funds in the market.