Overview
The table below summarizes our main datasets. It shows strategy returns for 207 trading strategies over different periods and for various alternative specifications beyond the original papers. You can download the table here. Note that most of the current data run through December 2021.
You can explore all available data via Google Drive File Explorer. You can find direct links to some featured datasets below.
Documentation
- The current data release is version 1.2.0 (created with the latest code release)
- See here for what has changed relative to the previous version
- Note: Most signals have not changed in-sample relative to the previous version. The R2 of regressing new signal return on old signal return is above .99 for the vast majority of signals. We discuss some exceptions (Accruals, Coskewness) in the release notes.
- Detailed info on the underlying papers and our implementations: Signal Documentation
- Includes acronyms, original paper (OP) references, and hand-collected results from the OPs
- Data release notes to help to get an overview of what is available
Featured Portfolio Return Datasets
- Monthly portfolio returns
- Monthly long-short returns of 207 predictors following OPs (wide csv)
- 207 csvs, each has returns for a portfolio sort using a different predictor following OPs (Gdrive folder)
- Use this link to get, say, 5 long portfolios based on B/M (BM.csv) or the long- and short portfolios based on dividend seasonality (DivSeason.csv)
- All 207 predictor portfolio sort csvs in a single file (still following OPs)
- Daily portfolio returns
Featured Stock-level Signal Datasets
- 204 predictive firm-level characteristics in wide format, signed so future mean returns increase in characteristics (1.6 GB zipped csv)
- Omits Price, Size, and STreversal, which can be downloaded from CRSP
- Code to automate the download available here
Previous data releases
See here for previous versions of our datasets (but you likely want to use the most recent one for your work)