Taiex Trend Volatility Strategies through Integration of Fuzzy Clustering and Grey Prediction

• Main Authors: Her-Chien Chao, 趙和謙
• Other Authors: Yen-Tseng Hsu
• Format: Others
• Language: en_US
• Published: 2005
• Online Access: http://ndltd.ncl.edu.tw/handle/40395538104717421895

碩士 === 國立臺灣科技大學 === 資訊工程系 === 93 === This research adopts techniques of data mining and artificial intelligence to build up analysis models from the historical data of stock market to assist investment decisions. The effective factors of stock market can be identified with Psychologicals, Fundamentals, and Technicals. This thesis primarily focuses on the technical analysis and refines the raw trend as the more reliable one — called as “PARK” by using filter rules learned from the past Taiex data. The risk level of each trading point in the PARK bull/bear sectors would be determined according to the statistics of price fluctuation and time duration, as well as the methods of fuzzy clustering and grey prediction. Finally, a series of feasible solutions with low risk and high profit would be developed by integrating various strategies for position control and then beat the market. The experimental results show three points: First of all, the performance based on PARK trend would be better than that on raw trend. Secondly, the more percentage the GMM-based short-term strategies occupy, the better performance the total strategies would obtain. Thirdly, the aggressive strategies would be better than that of the conservative ones if the trading signals are trusty.