Exploring the Relationship between Taiwan’s Stock Market and National Election with FCM Clustering, Fuzzy, and SVR Methods

• Main Authors: Yu, Ping-Chung, 余秉中
• Other Authors: Chiu, Den-Yiv
• Format: Others
• Language: zh-TW
• Published: 2011
• Online Access: http://ndltd.ncl.edu.tw/handle/15448522855137109390

碩士 === 中華大學 === 資訊管理學系碩士在職專班 === 99 === Stock market investment is always an esoteric knowledge especially in the economic environment with varied effective factors. Many scholars worked to identify the possible rules of the stock price in order to forecast the stock market by effective rules, then analyze the trading data and make strategies effectively and instantly by this way. Recent years artificial intelligence methods such as fuzzy theory and genetic algorithms are widely used to reduce the problems because the early methods like strategy tree suffered the problem to handle continuous data and neural networks has the over learning and local optimum problems. For the application of stock market forecast, combine several artificial intelligence methods can effectively reduce the defects and noise to improve the accuracy of forecast. Fuzzy theory and SVR are implemented to analyze the trading strategy of the Taiwan’s stock market during the period of national election in this study. The Fuzzy C-Means Clustering is used to build the membership functions of each characteristic for technical indexes at first. Then fuzzy indicators are created by fuzzy relation and defuzzification to generate the degree of trading strategies. The last step is to engage Support Vector Regression for exploring stock market behaviors. During the national election period, we found that the proposed model can perform better than buy-and-hold strategy according to the experimental results in this study. And the stock trading rules can be discovered by proposed model to gain better profit in the Taiwan’s stock markets during national election period.