Combination of Ensemble Algorithms with technical indicators to forecast stock market

• Main Authors: Chun-Teng Ko, 柯鈞騰
• Other Authors: Shian-Chang Huang
• Format: Others
• Language: zh-TW
• Published: 2013
• Online Access: http://ndltd.ncl.edu.tw/handle/39980944653314097713

碩士 === 國立彰化師範大學 === 企業管理學系 === 101 === In the recent years, many financial event impact the global stock market deeply. For instance, subprime mortgage crisis in USA, European sovereign debt crisis, the critical issue of the stock income tax in Taiwan etc., they make the global economic still heavily fluctuate. In empirical research, we can find the technical analysis is the useful tool to predict the movement of the stock price. Investigating the relations between volume and price variability in stock markets is the advantage of the technical analysis to forecast and control the trading rule. This paper adopted several technical analysis indicators, and use data mining approaches to sift out the results for trading signal. This paper not only experimented with single classifier such as Logistic, Support Vector Machine, and Neural Network, but also tested Ensemble Algorithms such as Adabooost.M1 and Bagging. This paper collected the daily data of the six countries which are USA, UK, Germany, Japan, Canada, and Taiwan, and examined the performance of the three single classifier and two ensemble algorithms. In conclusion, this paper found Logistic combined with AdaboostM1 have the best predictive.