Building a Stock News Sentiment Factor with Relevance Filters

Summary

This research summary explains a monthly stock-selection factor built from news sentiment. It combines the direction of news evaluations with their volume by summing sentiment scores for each stock over the month. The factor is refined by retaining news tagged as containing fundamental information and judged strongly relevant to the company.

The document reports that these filters improved the factor’s long-short and top-group results, including annualized returns, drawdown, benchmark-relative monthly win rate, and cross-sectional ranking behavior. It also describes variation by company size: effects were more persistent for larger stocks, while smaller stocks showed stronger reversal patterns and less influence from monthly news sentiment. Alternative measures include changes in sentiment, dispersion across scores, and the share of positive sentiment. These findings are presented as backtest summaries; the document does not provide the underlying study details here, and the reported relationships may not generalize across periods or markets.

Key ideas

• Monthly sentiment sums capture both news tone and the amount of coverage.
• Filtering for fundamental content and strong company relevance improved the reported factor results.
• The summary reports more persistent sentiment effects among larger stocks and stronger reversals among smaller ones.
• High positive sentiment consistency was associated with better subsequent returns in the described tests.
• The reported findings are backtest evidence and do not establish that the effects will persist.

Tags

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.