Stan constructs several News Sentiment signals using both the level and dispersion of Sentiment scores.
His results indicate that News Sentiment factors are effective in explaining the cross-section of stock returns. News sentiment signals are orthogonal to alternative proxies for Sentiment (e.g., IBES or short interest) and other systematic equity strategy factors used in Barra equity risk models (24 factors in total). The investment horizon for using the data is relatively short because the information ratio decay in the signal is fast.
MSCI uses RavenPack News Sentiment data to enhance Sentiment signal in the short-term trading Barra risk model.
Presentation held at the RavenPack 3rd Annual Research Symposium, New York, March 24th 2015.