Application of Neural Networks on Taiwan Stock Index Price Forecast and Trading Strategies

• Main Authors: We-De Li, 李威德
• Other Authors: Nen-Jing Chen
• Format: Others
• Language: zh-TW
• Published: 2003
• Online Access: http://ndltd.ncl.edu.tw/handle/16143114904589731184

碩士 === 輔仁大學 === 金融研究所 === 91 === This study applies Back-Propagation Neural Network to predict Taiwan stock index price. Disregarding technical and fundamental indicators, this thesis considers raw futures and cash index data as the inputs of the network. The Taiwan index futures opens earlier and closes later than its underlying cash market each by 15 minutes in a trading day. The daily opening futures index and previous trading day’s Taiwan cash and futures closing indices as well as some important American stock indices are used to predict the daily opening Taiwan cash index price. These indices are also used to construct a model to predict the daily trend of Taiwan stock index price. And finally, a speculative trading strategy is formed to evaluate the practicality of the model. Using the data from Jan.3, 2001 to Mar.31, 2003, the empirical results show that our model can predict the daily Taiwan opening cash index price and its trend satisfactorily. Besides, the trading strategy can earn a 10.58% annual return during the sample period.