Steve Y. Yang1, Qiang Song1, Sheung Yin Kevin Mo1, Kaushik Datta2, Anil Deane2

(1. Financial Engineering Program, Stevens Institute of Technology, USA;

2. Civil Finance Group, Northrop Grumman Corporation, USA)

 

 

Abstract: News sentiment has been empirically observed to have impact on financial market. However, finding a clear predictor of market returns using news sentiment remains a challenging task. This study investigates the relationship between news sentiment and cumulative market returns and volatility. We propose two methods for measuring the abnormal level of news sentiment, i.e., sentiment shocks and sentiment trend, and we analyze its relationship with market movements. The results show that abnormal levels of news sentiment are significant in predicting future market cumulative return and implied volatility of the S&P 500 index. Comparing the two methods, we find that the sentiment trend method demonstrates better performance than the sentiment shock method. In addition, our findings suggest that the strategy generated based on the abnormal news sentiment methods outperforms the buy-and-hold strategy through back-testing over the same time period.

 

 

Key words: news sentiment; financial market returns; market volatility; abnormal sentiment; trading strategy

 

JEL codes: G14, G17, G11, D83, D84