Promethean Markets
Log inGet started
Use cases

Built to be re-run.

Evidence-grade history earns its keep where results must survive being reproduced. Six ways teams put the archive to work — each mapped to the datasets that serve it.

Quant desks · prop traders

Backtesting & strategy research

Replay any certified day deterministically — the same bytes, the same manifest, every run. Order books on a 100ms grid mean your fill simulation sees what the market actually showed, and per-row freshness and lag fields tell you exactly what was knowable at each instant.

Polymarket Up/Down 100ms booksKalshi 100ms booksBinance trades & booksVenue trade tapes
Liquidity providers · venue teams

Market-making & liquidity calibration

Full depth on both sides, captured continuously, across venues that quote the same underlying. Measure spreads, depth decay, quote lifetimes and adverse selection against the reference exchange — then calibrate quoting models on windows where the outcome is already known.

Depth-20 books across venuesTop-of-book sweepsTrade tapes with aggressor sides
Researchers · arbitrage desks

Cross-venue price formation

The same question is priced on multiple venues at the same instant — a regulated US exchange, global CLOBs, an on-chain game — all against one clock. Study who leads, who lags, where probabilities disagree, and how fast disagreement closes.

All up/down venues on the shared 100ms gridThe settlement oracle streamBinance spot & futures reference
Risk teams · dispute research

Settlement & oracle forensics

Every Up/Down settlement can be reconstructed from four vantage points: the venue books, the trades, the reference exchange, and the exact oracle stream that decided it. When a resolution is contested, the archive holds the instant in question — with lag accounting.

Chainlink RTDS ticks + 100ms laneOn-chain rounds with lock/close pricesAligned venue books at settlement
AI labs · model teams

LLM & ML training, evals, calibration

Resolved markets are labeled ground truth with precise timestamps — the raw material for leak-free forecasting evals, calibration studies, and reward signals. Freshness fields make honest cutoffs enforceable by construction.

Resolved windows with settlementsTimestamped probability pathsManifest-pinned reproducible snapshots
See the LLM implementations guide →
Universities · analysts

Academic research & journalism

Citable and reproducible: every file carries a SHA256 that reviewers can check, corrections live in a public ledger, and coverage claims are verifiable against the same index the API serves. Your methodology section writes itself.

The full archive with manifestsThe corrections ledgerThe public availability index

Start with a free key, or tell us the coverage you need.

The full catalog and schema docs are open — integrate first, buy when it fits.

Get startedBrowse the datasets