Using Analyst Forecast Revisions as Equity Selection Factors

Summary

This study examines equity factors built from changes in analyst forecasts and tests how forecast stability affects their relationship with next-month stock returns. It reports that revision factors are positively associated with subsequent returns, but their selection performance depends on the lookback period. Short windows leave many unchanged forecasts in the sample; longer windows improve coverage, while very long windows can weaken the signal because the information becomes less timely.

The analysis also groups stocks by the historical volatility of their forecasts and applies time-series standardization to revisions. The document reports more stable results after standardization, with stronger information coefficients and information ratios, including after industry and style adjustments. These are reported study findings, not guarantees of future performance.

A key data caveat is that simulated or carried-forward forecast records are less reliable than recent analyst forecasts. The study finds little selection value in those less reliable subsets and reports better results when missing or unreliable observations are assigned zero. The supplied text is an abstract rather than the full research paper, so it does not provide enough detail to assess the full methodology, sample, or implementation.

Key ideas

• Analyst forecast revisions are reported to have a positive relationship with next-month equity returns.
• The lookback period affects both coverage and signal timeliness, producing a reported rise and then decline in rank information coefficient.
• Grouping stocks by the historical volatility of their forecasts can improve revision-factor selection results.
• Time-series standardization is reported to make the factor more stable and improve its information ratio.
• Simulated or carried-forward forecast records show little selection value, and zero-filling unreliable observations is reported as a useful alternative.

Tags

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