|
Compared with the past, as the proportion of the transaction amount of the institutional investors in the Taiwan stock market has been increasing, the trading trend of institutional investors now seems able to dominate the ups and downs of both the market and individual stocks. This study uses the stock of TSMC as the research object, and applies the Deep Neural Network's algorithm (DNN). In addition to the practical technical-analysis indicators like RSI, KD, MACD and BIAS, it also looks at the information provided by institutional investors in order to specify the influences on the short-term increases in value of TSMC stock. The study shows that, regarding the technical analysis indicators, the BIAS is more related to the returns, while with the institutional investors, the net buy/sell of foreign investors and the investment trusts have more significant influence on the rate of stock return. It seems to have no significant impact on the odds whether we use a regression model or DNN with the trading strategy set in this study. However, with both models we have better odds after incorporating the significant variables of institutional investors, than if we are only using technical analysis indicators. Therefore, we can say that the net buy/sell of foreign investors and investment trusts both have an impact on the returns with TSMC stock. By using the trading strategy proposed by this study, the rate of return obtained during the backtesting period is higher than that of the Taiwan Weighted Stock Index. it shows that this trading strategy is a feasible one, at least within certain limits, and that it can provide a sound basis for judgments regarding investments in the future.
|