Peter Hafez, director of Quantitative Research, RavenPack
In this study, a set of indicators is created that capture abnormal sentiment for different news topics as they relate to a country or economy. Indicators are tested in long-only and long/short strategies on the S&P 500 and find they outperform traditional price-driven models.
Specifically, findings show:
- Since September 2007, both news-based strategies have Information Ratios (IRs) of 1.45 versus a buy-and-hold S&P 500 strategy IR of 0.23 and a pure return-driven model IR of 0.73
- Maximum drawdown in news-based models is reduced by over 70% on the S&P 500 compared to a buy & hold strategy
- Abnormality in aggregate corporate news is a key driver of outperformance