Neural Network Portfolio Construction of TSEC Taiwan 50 Index：A Comparison between Technical Analysis and Fundamental Analysis

• Main Authors: Chuan-Cheng Hsieh, 謝泉澄
• Other Authors: Chin-Sheng Huang
• Format: Others
• Language: zh-TW
• Published: 2010
• Online Access: http://ndltd.ncl.edu.tw/handle/65697140040684290750

碩士 === 雲林科技大學 === 財務金融系碩士班 === 98 === The study separately utilizes neural network models of technical indicators and fundamental variables forecasting the constituents stocks of TWSE Taiwan 50 index and comparing the performance of the constructed portfolios in terms of benchmarking TWSE TAIEX and TWSE Taiwan 50 index. Empirical results indicate that prediction accuracy of neural networks of technical indicators and fundamental model respectively account for 60.11% and 72.86%. Moreover, the evidence also shows that the neural network portfolios models of the predicted top 5, top 10, and top 15 outperform the benchmarks of TWSE TAIEX and TWSE Taiwan 50 index. Investors can benefit from neural networks prediction in investment decisions.