Chess Multiverse Research
Last Updated: Saturday, January 24, 2026

Open Data Repository

Curated, clean, and annotated chess datasets for research. All datasets are hosted on Zenodo to ensure long-term preservation and distinct DOIs.

Sort by:

Blitz Decision Fatigue (v1.0) Published

DOI: 10.5281/zenodo.xxxxxxx
PGN 1.2 GB (Large) Jan 2026 CC BY-SA 4.0
A clean subset of 50,000 anonymized Blitz games filtered for players rated 1500-2200. Specifically annotated with move-times to calculate decision speed degradation over game phases.
Fields: move_ply, clock_time, time_spent, eval_centipawn, is_blunder

Lichess Open Database (Mirror) External Mirror

Monthly Updates Various Sizes
The gold standard for open chess data. We link to the "Elite" subsets which are most useful for researchers looking for high-quality play patterns without engine noise.

Tactical Error Clusters (KID) In Preparation

Planned DOI: Pending Review
JSON ~450 MB (Medium) Q3 2026
A proposed dataset mapping 12,000 instances of typical tactical blunders in the King's Indian Defense structure. Designed specifically for training pattern recognition models.
Target Fields: fen, blunder_move, correct_move, tactical_motif_tag

How to Cite

Please cite the specific dataset DOI when using our data in your research.

Varshney, S. (2026). Blitz Decision Fatigue Dataset (v1.0). Chess Multiverse. DOI: 10.5281/zenodo.xxxxxxx
Data Ethics: All datasets are anonymized to remove PII and comply with platform terms of service.

Have data to share?

We host academic chess datasets for free under Creative Commons licenses.

Contribute to Repository →