Summary: | 碩士 === 國立臺灣科技大學 === 資訊工程系 === 99 === A mutual fund is a bundle of investments (whether stocks, bonds, money, art, precious metals). These investments are managed professionally and shares of the mutual fund are sold to the public. This allows mutual fund holders to diversify among many companies instead of owning a single stock. Mutual funds have become the most popular products for diversity of investment. Traditionally, people are used to analyze the historical data and the market manually, however the return is not convincing most of the time and the risk factor is very high. Hence, a successful trading strategy is necessary to achieve the profit and good market forecast. In this thesis, an efficient and simple trading strategy model is designed based on optimization algorithm, Turbulent Particle Swarm Optimization (TPSO) in combination with technical indicators namely Moving Average Convergence-Divergence (MACD). To check the stability and performance of the proposed technique, different window sizes (different time periods) of training data are used. From the experimental finding, it turns out that proper duration of training period is very important to achieve better profit and seven years training period gives the best performance in comparison with other window sizes. The performance of each fund on average has been improved more than the original about 38% and 22% for 7 and 8 years training period respectively in testing phase.
|