JKP Global Factor Data

This is the dataset behind Jensen, Kelly and Pedersen’s Is There a Replication Crisis in Finance? (Journal of Finance 78(5):2465–2518, 2023). It supplies characteristic-managed long-short factor returns and the underlying sorted portfolios — 153 characteristics grouped into 13 themes, for 93 countries and 4 regions per the site — at monthly and daily frequency, in capped value-weighted, value-weighted and equal-weighted variants across all-stock, mega, large, small, micro and nano-cap universes, with industry returns (GICS and Fama-French 49) and NYSE size cutoffs alongside. The machine-readable availability manifest lists 102 country/region keys for factors, because it also carries aggregates such as world, developed and emerging. Coverage dates above are for the US market factor series.

Access is two-tier and the distinction matters. Factor returns, sorted portfolios, industry returns and cutoff files download from a public S3 bucket with one unauthenticated GET and no registration. Stock-level firm characteristics do not: building those requires the open jkp-data pipeline plus a WRDS subscription with CRSP and Compustat entitlements, and the README is explicit that the authors do not distribute WRDS, CRSP, Compustat or IBES data, pointing non-subscribers to the precomputed factor portfolios instead. One gotcha: download URLs embed literal square brackets that must be percent-encoded, and a malformed request returns HTTP 403 from S3 rather than 404.

License: the split is real. The pipeline code is MIT, but the data is CC BY-NC 4.0 — commercial use of the distributed series is not permitted. The site also runs a Common Task Framework in which entrants submit portfolio-construction models scored on Sharpe ratio, return, volatility and maximum drawdown over a 1990–2023 test period, with a public leaderboard of completed external submissions; those are harness backtest figures rather than live trading results, and the CTF tables sit behind the same WRDS requirement.