Open:FactSet Forum

Measuring Alpha of Stocksnips News Sentiment Signal

Stocksnips collaborated with CMU researchers to measure the Alpha potential of the News Sentiment signal. A Sentiment dataset for 984 US Equities was provided for the period 2017/01 – 2019/06. The method used was to rebalance the portfolio daily and measuring returns using a Long /Short strategy (L/S) for Value Weighted (VW) and Equal Weighted (EW) portfolios. The top 50 ranked tickers would be long (buy) and the bottom 50 ranked tickers would be short (Sell) and a Net Zero method was used for the daily portfolio amount. The portfolio returns were regressed on Fama-French factors to arrive at the Alpha attributable to News Sentiment. For details please see attached reportMeasuring Alpha of Stocksnips Sentiment Signal.pdf (227.8 KB)