Quantamental Investing with Sentiment Data
The author shows how a sentiment signal, derived from news and social media, performs as an overlay to an existing leveraged buyout strategy (LBO). The study should be of interest to activists, risk arbitrageurs and speculators on the long side as well as market participants (typically quants), looking to eliminate event risk on the short side.
The QMIT Leveraged Buyout (LBO) Model
The paper introduces the QMIT LBO model developed by Milind and his team. In addition to a long-term hit rate of 41%, the Top 100 model predictions can be traded profitably as an equal-weighted long portfolio with a Sortino ratio of close to 2.5 over the 19-year backtesting history.
Enhancement via Sentiment-Based Alternative Data
To address potential event risk, a sentiment signal is introduced as an overlay into the model. In particular, they apply RavenPack’s Sum Excess Sentiment Indicator (SESI). Given that SESI captures news-based sentiment, which may include rumors on LBO names, it is logical to ascertain whether benefits may accrue from trading a combined quantamental signal.
Key Results of this Quantamental Strategy
Considering the weekly rebalanced LBO Top 100 signal with a SESI overlay, the authors find the following key results:
- Absolute +8.6% improvement to annualized returns, for the best overlay scenario tested
- +46% boost to the Sharpe ratio, for the best overlay scenario tested
- Substantial improvements over the benchmark signal – in returns, Sharpe, and Sortino
- Significantly improved drawdown profile of the overlaid strategy