Historical Polymarket Sports Data

Sports markets have a clear event clock. Pre-game information, lineup news, the start of play, scoring events, interruptions, and the final result can each create a different liquidity regime. Tick-level records let you divide a market into those phases instead of treating the entire event as one price series. You can then measure how quickly probabilities changed, whether spreads widened around uncertain moments, and how much depth was available when a signal would have traded.

The order book and trade datasets answer different questions. L2 snapshots and updates show quoted liquidity, queue shape, and the prices that were available to a strategy. Trades show where executions actually occurred. On-chain fills provide a separate view of settled activity. Combining the three helps distinguish a temporary quote from executable interest and makes it possible to build backtests that account for spread, depth, and event timing.

PMData files contain Polymarket market activity, not official scores, schedules, or sportsbook odds. For event studies, join the market slug and timestamps to the external sports source you trust, normalize all timestamps to UTC, and define the information cutoff before evaluating a signal. Keeping those inputs separate reduces look-ahead bias and makes comparisons across leagues or event types easier to reproduce.