| Summary: | 碩士 === 國立雲林科技大學 === 財務金融系 === 107 === Thanks to the rapid development of the Internet, investors can now receive information in a more rapid way. However, it also determining which are the useful ones became more complicated. Therefore, how to correctly interpret the news in the market to seek profits is an important issue for investors in the market today.
This study uses media information as the proxy variable of investors sentiment and classify the emotions by the content of the news in two categories, and also subdivide the strength and weakness by the affirmation of the content.
This article uses the Internet Finance News of the Commercial Times as the main study sample and the JebaR package in the R language. We also created a dictionary for affirmative words and used Chen Hongming (2017) and Lin Yixuan’s (2013) sentiment words dictionary to compare the news content, and thereby converting the text into the sentiment ratio and the sure ratio. After testing the return rate of change and volatility rate with additional control variables by OLS, ARIMAX, ARIMAX-GARCH and Quantile Regression, we resulted in with the following main findings:
(1) After adding two investor sentiment variables, the ability to predict the return rate of change and volatility rate of the next-day TAIEX and the FITX has a significant improvement effect.
(2) The sentiment ratio has a positive effect on the return rate of change in the TAIEX on the next day and the sure ratio has a negative effect in the two depending variables.
(3) The sentiment ratio has a negative effect and the sure ratio has a positive effect on the volatility of the two rate returns of change on the next day.
(4) The sure ratio has an asymmetric effect on the return rate of change and the volatility of the two rate returns of change on the next day.
|
|---|
