The Effect of Online Comments on Stock Trends by Sentiment Analysis
This paper is to examine whether online stock commeuts could affect the stock trend by constructing ARMA-GARCHX model and ARMAX-GARCH nodel.The relationship between the sentiment of investors toward certain stocks and the stock trend is expected to be found for enabling better investment decisious.Our model is featured with three advantages: bigger sample size, better timeliness, and closer to true emotion of investors.The investors' sentiment is set as an external variable, which is then combined into the two proposed models.The empirical study is conducted with online comments on the 30 selected stocks of CSI 300 Index.The result shows that ARMAX-GARCH model outperforms ARMAGARCHX model in term of prediction accuracy.In most situations, however, ARMAX-GARCH model prefer current sentiments to previous ones.By contrast, ARMA-GARCHX model uses more past sentiments, thus being more applicable to stock forecast.
online stock commentssentiment indexstock trendARMA-GARCHX modelARMAX-GARCH model
NETL
NSTL
万方数据
维普
