Building Equity News Sentiment Factors from Relevant Fundamental Coverage
Summary
The document describes monthly stock-selection factors built from news sentiment. Summing a stock’s monthly article sentiment scores combines the direction of coverage with its volume, while averaging scores captures sentiment without the same direct emphasis on article count. The proposed refinements filter for articles tagged as containing fundamental information and strongly related to the company.
The document reports that these filters improved the long-short factor’s annualized return and the top group’s return, drawdown, benchmark-relative monthly win rate, and performance monotonicity. It also describes mixed results across market capitalizations: sentiment appears more persistent for larger stocks, while smaller stocks show stronger reversal and weaker results for some smaller-float names. Alternative measures include monthly sentiment change, dispersion in scores, and the share of positive scores. The reported findings are backtest evidence; the document gives no details here about sample period, costs, or robustness, so they do not establish that the effects will persist.
Key ideas
- Summing monthly news sentiment scores combines sentiment direction with coverage volume.
- Filtering for strongly related news with fundamental content reportedly improves factor backtest results.
- Monthly sentiment appears more persistent for larger stocks, while smaller stocks show greater reversal.
- Sentiment change and score dispersion do not show clear monotonic sorting in the described tests.
- A high total sentiment score with low dispersion, or a larger share of positive scores, is associated with stronger subsequent returns in the reported analysis.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.