New: the v3.1 preview. Faster queries and smaller downloads, three sample hours to test. Tell us whether it is ready to roll out. Try the preview
Polymarket Orderbook Archive archive.pendulumflow.com

How to cite

If this data is in your work, cite the archive and say which era you used. The licence is CC BY 4.0: the one thing it asks is that you say where the data came from.

Plain text

pendulumflow. Polymarket Orderbook Archive. https://archive.pendulumflow.com, accessed 2026-10-21. CC BY 4.0.

BibTeX

@misc{pendulumflow_polymarket_orderbook_archive,
  author       = {pendulumflow},
  title        = {Polymarket Orderbook Archive},
  year         = {2026},
  howpublished = {\url{https://archive.pendulumflow.com}},
  note         = {Accessed 2026-10-21. Hourly order book event streams in Parquet, CC BY 4.0.}
}

Citation File Format

The repository carries a CITATION.cff, which GitHub shows as "Cite this repository" and citation managers read directly.

A DOI

No DOI yet. Until there is one, cite the date you fetched the data and keep the checksums you downloaded with: the archive grows by an hour every hour, and an older hour can be rebuilt to a better state later (each hour's audit report lists its history), so the checksums are what name the exact bytes you used.

Research using this archive

None yet that we know of. If your work uses this data, tell us and we will list it.

If you have used this data, add yourself: send a title, authors as you write them, a year, a link a reader can open, and one line saying which era or hours you used through the contact details on the credits page, and we will list it. The eras are not interchangeable, so that last line is what makes the list worth reading.

For a machine

A Croissant description of the archive is served at /croissant.json. It names every column and its measured type, so a loader can open the files without reading this site first. Two things it cannot say in its own structure, and says in prose instead: each corpus's columns were read from the footer of a single named hour, and the number of columns is not the same across the archive.

The mirrored eras have their own collectors

The V1 and V2 eras under pmxt/ are the work of PMXT and are mirrored here unmodified under CC BY 4.0; the AG6 V2 archive under third-party/ag6/ is the work of AG6, who publish no licence terms. If your work rests on one of those eras, credit its collector as well, and see the credits for who they are.

Credit the collector: pendulumflow for V3, PMXT for V1 and V2, AG6 for their V2 archive. Ours and PMXT's are CC BY 4.0; AG6 states no licence. How to cite.

Serving these bytes is not endorsing them. We are not affiliated with Polymarket.

For AI readers: llms.txt, what this archive holds and the questions it cannot answer.

https://x.com/PendulumFlow

Join the Pendulum Flow Discord: other people who build on prediction-market data, comparing notes and sharing tooling and findings.