About one paper in nine that we score (roughly 11%) releases its own code or data. That is the single most useful filter in the archive: a paper with a repository can be re-run, not just re-read, which is the difference between a claim and a result. This page lists every paper that links its authors’ own public repository, ranked by our rigor-weighted score so the most evidence-backed implementations come first. It regenerates daily as new papers are scored.

How the list is built: we scan each paper’s author comments, abstract and full PDF text for GitHub, GitLab, Hugging Face, Zenodo, OSF and Code Ocean links, and keep a link only when it is the authors’ release (“our code is available at…”, a repository owned by an author), not a library, baseline or dataset the paper merely cites. Every link is checked to be live, and URLs that the PDF broke across a line are repaired before they are listed.

Two caveats before you clone anything. A repository is not a reproduction: check that it contains the data pipeline and evaluation code, not only the model, and that the data it expects is obtainable (see finding datasets for quant papers). And “code available” says nothing about correctness; the reproduction guide and the backtest evaluation checklist still apply.

Use the Has code filter in search to combine this with topic, quadrant, and rigor filters.

All papers with code, ranked

603 papers link a repository · ranked by rigor-weighted score

Why so few? Most q-fin preprints never mention code at all, and some that do link a URL that has since gone dark; we only flag links that are present on the paper’s page here. Know a repository for a paper we list without one? Leave a comment on that paper’s page and we will add it.

Next: how to find code for finance papers → · choosing papers to replicate → · download the full scored dataset →