Polymarket Resolution Typology Dataset

A three-class classification of 911,237 Polymarket prediction markets by resolution mechanism, released as a methodological resource for research on prediction-market structure and informed-trading detection. Companion dataset for ForesightFlow: Real-Time Detection of Informed Trading in Decentralized Prediction Markets.

Why this dataset exists

Applying information-based microstructure models — in particular, computing the Information Leakage Score (ILS) — requires knowing how each market resolves. Deadline-resolved markets behave structurally differently from event-resolved markets and require different reference timestamps, baseline assumptions, and interpretations of pre-event price drift. No public classification existed prior to this release.

Coverage

Resolution type	Count	Share
unclassifiable	853,774	93.7%
deadline_resolved	56,318	6.2%
event_resolved	1,145	0.1%

Category	Count	Share
other	771,424	84.7%
regulatory_decision	71,588	7.9%
military_geopolitics	47,580	5.2%
corporate_disclosure	20,645	2.3%

Total: 911,237 markets · ~$59.8 B USDC total volume · date range 2020-10-02 — 2026-04-26.

Format

Available as gzipped JSONL and Parquet. Each record contains: condition_id, question, resolution_type, category_fflow, created_at, resolved_at, volume_usdc, outcome.

Intended use

• ILS-based informed-trading analysis requiring per-market resolution type
• Studies of crowd forecasting accuracy, calibration, and deadline effects on liquidity
• Market design research differentiating deadline-resolved from event-resolved venues

@misc{nechepurenko2026prt,
  title     = {Polymarket Resolution Typology Dataset},
  author    = {Nechepurenko, Maksym},
  year      = {2026},
  publisher = {ForesightFlow / Devnull FZCO},
  url       = {https://github.com/ForesightFlow/datasets/tree/main/polymarket-resolution-typology},
  note      = {Version 1.0, CC-BY-4.0. Snapshot cutoff: 2026-04-27}
}