Combo & Parlay Archive: Polymarket + Kalshi

Polymarket + Kalshi combo and parlay data, Aug–Oct 2026: fills, prints, decoded legs and outcomes in Parquet, with a derived coverage map.

Full schema & known gaps →· Sold as-is — but if a file is broken or won't open, email [email protected] and we'll send a fixed file or refund you. No hassle.

Inspect the free sample rows and schemas before buying.

Frozen artifact delivery: the product page and shipped documentation state the exact coverage, row counts and known limitations. A purchase delivers the advertised frozen artifact; no future refresh is promised unless the checkout terms explicitly say so.

Mixed competition combo records Observed partial coverage

Combo legs span multiple competitions and seasons. Individual event-to-season mapping and a complete fixture denominator are unverified; capture is partial.

Polymarket order-book history is not included. Kalshi includes one top-of-book snapshot fetched at registration per combo, not a repeated book series. Exact registration-snapshot time bounds are not recorded in this metadata.

Kalshi trade exchange times:
2026-08-06T03:11:29.651294Z → 2026-10-17T18:11:31.361111Z
Polymarket fill block times:
2026-08-06T03:08:35Z → 2026-10-17T23:59:57Z
Artifact inclusion dates:
2026-08-06 → 2026-10-17

UTC timestamps retain their recorded precision; dates alone identify UTC days. Fill/trade event times, recorder retrieval times and artifact inclusion dates are separate. These clocks do not describe an order-book history.

Event and retrieval times by source
kalshi_combo_trades
Kalshi trade exchange times:
2026-08-06T03:11:29.651294Z → 2026-10-17T18:11:31.361111Z
Kalshi trade retrieval times:
2026-08-06T03:11:39.782549Z → 2026-10-17T18:14:41.584760Z
created_time is the exchange trade time; captured_at is recorder retrieval time. captured_clock_fault flags the documented recorder-clock rollback.
polymarket_combo_fills
Polymarket fill block times:
2026-08-06T03:08:35Z → 2026-10-17T23:59:57Z
Polymarket fill retrieval times:
2026-08-06T03:28:19.459801Z → 2026-10-18T00:13:20.860224Z
block_time_source distinguishes exact block-header times from interpolation or extrapolation. Retrieval clocks are provenance; captured_clock_fault flags the documented recorder-clock rollback.
Scheduled sports event times:
start unknown → end unknown

Scheduled event times are not inferred from capture windows or combo trade dates.

Coverage qualifications
  • Combos mix legs from many sports, competitions and seasons; no season scope is declared.
  • No season scope declared in the archive manifest
  • Complete-season fixture denominator has not been verified

What this archive adds

Polymarket combo fills are public on-chain events: anyone can re-read them from Polygon. Kalshi documents per-market trade-history endpoints; inspect their current pagination, retention and availability before treating a downloaded history as complete. What this archive adds is the decoding and joining: decoded Polymarket combo legs, names where available and recorded on-chain outcomes; Kalshi's combo definitions with leg labels, combo and leg results, and the book snapshot our recorder took when it registered each combo, in one schema with a derived coverage map and a documented capture-gap map (gaps.csv).

What's in the dataset

Dated snapshot: 1,628,116 Polymarket combo fills · 375,291 Polymarket combos · 1,570,872 Kalshi combo definitions · 2,062,683 sampled Kalshi combo prints · 20,633,854 total rows · Kalshi trade prints are a small sample: about 2.0% of all Kalshi combo trades Aug 6-Oct 17, 2026, measured against Kalshi's own history · 2026-08-06 through 2026-10-17 · artifact vintage 2026-10-17; separate from observation times

Research uses for this dataset

Cross-venue comparison

Compare combo structure, prices and settlement between Polymarket and Kalshi over the same weeks.

Data hygiene first

Use the derived recorder-gap flags in gaps.csv alongside days.csv and the source timestamps. These files describe measured capture gaps; they do not establish that every outage is known or that no trade occurred in a missing window.

See the real rows before you buy

We publish real sample rows — not a marketing mockup. Load them in DuckDB, inspect the schema and coverage contract, then decide.

← All data archives · Backtesting Polymarket strategies — tools & datasets · Data license