| Summary: | 碩士 === 國立雲林科技大學 === 企業管理技術研究所 === 87 === This study uses genetic algorithm (GA) to search the technical rules with best trading profits and their combining strategies. The GA technical trading rules are tested by the daily trading data of Taiwan Stock Exchange Index (TSE Index). Then we examine whether the GA technical trading rules are better than Buy and hold Strategy. The technical indicators of this study include moving average, relative strength index, and moving average of volume. By combining single indicators, we construct seven different trading rules, namely, single, double, and triple indicator rules. According to market sentiment, we use three sample data: 1984-1989 of bull market, 1990-1993 of bear market, and 1991-1996 of cyclical market. The train periods of these samples are 4 years, 2 years, and 4 years respectively. Two-year testing periods are used to compare the commutative return of GA technical rules to the TSE Index. We found the cumulative returns of GA technical trading rules are all higher than TSE Index in all training periods. However, the returns of testing periods are different in each testing period. The 1988-1989 of bull market, only two GA technical trading rules exceed TSE Index. On average, GA technical trading rules are not significantly better than Buy and hold strategy in this period. In 1990-1991 of bear market, six GA technical trading rules ate higher than buy and hold strategy and have excess returns. In 1995-1996 of cyclical market, the trading profits of GA technical rules are less than the TSE Index and show no excess returns.
|
|---|
