Summary: | 碩士 === 國立成功大學 === 統計學系碩博士班 === 100 === The high returns is the benefit in stock markets but associated with high risk. To control the risk, the investment analysis is required. Technical analysis is a useful tool to provide the signals to enter or exist a trade by various technical indicators calculated from the historical transaction information. However, there are some restrictions on the usage of every indicator. To avoid the mistakes in deciding the timing of trading, considered multiple indicators is recommended. But how many the indicators would be enough in judging the timing? In this study, an integrated time-choosing investment strategy is proposed.
The Taiwan Top 50 ETF(Code:0050)is illustrated as a target in this study. Apply factor analysis to extract the common factors from bunches of technical indicators. Then build a discriminant function for the trend of price with cluster analysis and discriminant analysis. Also, use the common factors to construct trading decision model with back-propagation network and dynamic training method. Finally, the balance of margin, the net buy/sell of Institutional investors and basis in futures market are joined together to provide a trading strategy. The result shows, with the method in this study, can capture more short-selling time points in a bear market and get a higher rate of return than the benchmark.
|