Title: Using large and heterogeneous sources of sentiment and attention data for predicting stock market volatility
Authors: Daniele Ballinari - University of St Gallen (Switzerland)
Francesco Audrino - University of St Gallen (Switzerland)
Fabio Sigrist - Lucerne University of Applied Sciences (Switzerland) [presenting]
Abstract: The impact of sentiment and attention variables on volatility is analyzed by using a novel and extensive dataset that combines social media, news article, information consumption, and search engine data. Applying a state-of-the-art sentiment classification technique, we investigate whether sentiment and attention variables contain additional predictive power for realized volatility when controlling for a wide range of economic and financial predictors. Using a penalized regression framework, we identify investors' attention, as measured by Google searches about financial keywords (e.g. ``financial market'' and ``stock market''), and the daily volumes of company-specific messages posted on StockTwits to be the most relevant variables. In addition, it is shown that attention and sentiment variables are able to significantly improve volatility forecasts, although the improvements are of relatively small magnitude from an economic point of view.