The relative importance of overnight sentiment versus trading-hour sentiment in volatility forecasting

We examine the relative importance of overnight sentiment versus trading-hour sentiment in forecasting volatility. Previous studies on investor sentiment either ignore overnight sentiment or aggregate overnight sentiment with trading-hour sentiment. With the help of Chinese sentiment dictionary, we extract investor sentiment from Chinese internet social forums. Our empirical analyses suggest conclusively that investor sentiment significantly affects volatility. In particular, overnight sentiment is more informative than trading-hour sentiment in forecasting volatility, and has higher predictive power than overnight returns, which are widely used to capture overnight information. Our results hold in a series of robustness tests, including in highly volatile subsample, alternative rolling window size, and alternative sentiment proxy.