ForesightFlow
← Datasets
polymarket-tnews-tevent-recovery-v1/v1.0 · CC-BY-4.0

Polymarket T_news / T_event Recovery

Public-event and news-arrival timestamps for 2,052 resolved Polymarket markets, recovered across three methodological tiers: UMA Oracle proposer evidence (Tier 1, n=12, confidence 0.95), GDELT proxy (Tier 2, n=1,993, confidence 0.60), and LLM-assisted multi-source verification (Tier 3, n=47, confidence 0.80–0.90). Snapshot: 2022-12–2026-04.

Event timestamps for resolved Polymarket markets are not available in any standard public form. This dataset releases the consolidated recovery record for 2,052 markets so that downstream work on event studies, information-leakage analysis, and price discovery can reuse these anchors directly rather than re-deriving them from scratch.

Coverage

TierMethodnConfidence
Tier 1 — UMA Oracle proposer evidenceuma_proposer_evidence120.95
Tier 2 — GDELT / resolved-at proxygdelt_keyword_match + 24h offset1,9930.60
Tier 3 — LLM-assisted multi-sourceclaude-haiku-4-5 + web_search470.80–0.90
Total2,052

Category labels reflect the v1.1 taxonomy correction (esports reclassified out of military_geopolitics).

Tier guidance

  • Tier 1 (n=12) — gold standard; use as ground truth for benchmarking.
  • Tier 3 (n=47) — LLM-verified with source citations; use for case studies where provenance matters.
  • Tier 2 (n=1,993) — structural proxy (t_resolve − 24h); adequate for population-level analysis, not for individual market studies where the exact event time matters.

Files

FileDescription
data/tnews_tevent_recovery_v1.jsonlFull dataset (2,052 records)
data/tnews_tevent_recovery_v1.csvSame data, CSV format
data/tier1_uma_subset.jsonlTier-1 subset only (12 records)
data/tier3_llm_subset.jsonlTier-3 subset with source URLs (47 records)

Quick start

import json

with open("data/tnews_tevent_recovery_v1.jsonl") as f:
    records = [json.loads(line) for line in f]

# High-confidence Tier 3 only
tier3 = [r for r in records if r["tier"] == "tier3_llm"]
print(f"Tier 3: {len(tier3)} markets")

Citation

@misc{nechepurenko2026tnews-tevent-dataset,
  title     = {Polymarket T\_news / T\_event Recovery Dataset},
  author    = {Nechepurenko, Maksym},
  year      = {2026},
  publisher = {ForesightFlow / Devnull FZCO},
  url       = {https://github.com/ForesightFlow/datasets/tree/main/polymarket-tnews-tevent-recovery},
  note      = {Version 1.0, CC-BY-4.0. Accompanies: Empirical Evaluation of Deadline-Resolved Information Leakage on Documented Polymarket Insider Cases (arXiv:2605.02286)}
}