The Analysis of Investing Strategies with Text Mining

• Main Authors: LIN, ZHENG-XIU, 林政修
• Other Authors: LIN, SHIN-HUNG
• Format: Others
• Language: zh-TW
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/8taqjf

碩士 === 國立雲林科技大學 === 財務金融系 === 105 === This study analyses online financial news about four companies collected from Chinatimes.com, cnYes and NOWnews for about two years, as a data set to predict their stock prices. These four companies are Taiwan Semiconductor Manufacturing Company Limited, Hon Hai Precision Industry Company Limited, HTC Corporation, and LARGAN Precision Company Limited. Text mining technology is applied to transfer online news from unstructured data into structured numerical data. Online news is analyzed automatically to classify document and predict up and down trends of stock prices through this design. Furthermore, a simulation of stock trading strategy is conducted by the model to predict trends of stock prices. The accuracy of model prediction and revenues of the stock trading strategy are both monitored. Through this design, this study finds empirical data to support the idea that text mining model is better than technical analysis as stock trading strategy when companies face a phase of transition or significant events.