Open Source Asset Pricing

Open Source Asset Pricing reconstructs each published predictor from its original paper, drawing on CRSP, Compustat, IBES, OptionMetrics and 13F inputs. The project’s own registry, SignalDoc.csv, holds 331 rows — 212 categorized “Predictor”, 114 “Placebo” and 5 “Drop” — and records each signal’s authors, year, journal, sample period and economic category. By underlying data source the 331 split as Accounting 196, Price 56, Analyst 21, Trading 20, Other 13, Options 9, 13F 8 and Event 8. Release 2.0.0 (2025-10-22) provides monthly and daily portfolio returns for the 212 predictors under several construction methods, plus 209 stock-level firm characteristics; most series run through December 2024, while option-implied volatility predictors end December 2022.

The data is free and needs no registration. A Python package (pip install openassetpricing) exposes dl_signal_doc(), list_port() and dl_port() returning pandas or polars dataframes, with vintage releases selectable by date; an R package covers the same ground. Note that the releases themselves live in Google Drive folders hard-coded in the package’s urls.py, so availability depends on those folders staying shared — there is no versioned DOI or institutional archive. The package’s own source repository carries no LICENSE file, although its PyPI metadata declares GPLv2.

The GPL-2.0 above was read from the LICENSE file of the construction-code repository. No separate license is published for the data itself; the only rights-adjacent text on the data page is a disclaimer that the views expressed are the authors’ own and do not necessarily reflect the position of the Board of Governors of the Federal Reserve or the Federal Reserve System (Andrew Y. Chen is at the Federal Reserve Board). The README asks users to cite the method paper, Open Source Cross-Sectional Asset Pricing, Critical Finance Review 11:207–264 (2022).