- NBA ·
NFL · EPL · FIFAMarkets - ~2 Hz sportsSnapshot rate
- Pre-game +
liveWindow - Scale +
Tier
A sports market is a probability with a clock running against it. In the hours before kickoff it drifts on team news, weather, and flow; once play starts it swings on every score and momentum shift; and at the final whistle it resolves. With sports snapshots captured at roughly 2 Hz, you can record that whole arc, the pre-game drift and the in-game swings, as one continuous order-book history.
Sports data sits on the Scale tier and above. The capture rate is roughly 2 Hz, slower than the ~20 Hz on crypto, because the underlying reprices less furiously most of the time, which is plenty to trace pre-game line moves and to catch the step changes when a goal goes in or a quarter turns. As always, settlement comes from Polymarket; we capture the book around the outcome, not the scoreboard itself.
Three phases of one market
Pre-game drift
In the run-up to kickoff the probability eases on team news, lineups, and flow, a slow line move you can trace snapshot by snapshot.
- Lineup and news
- Gradual repricing
- Line into kickoff
In-game swings
Once play starts every score and momentum shift jolts the book, sharp steps in the implied probability, then settling.
- Step on each score
- Momentum repricing
- Sharp then settle
Toward resolution
As the result firms the probability presses toward one or zero, the spread tightening or gapping as makers manage the endgame.
- Pressing to 0/1
- Endgame spread
- Settles at the whistle
The fields the arc is built from
The same snapshot schema that serves crypto serves sports, so the workflow carries straight over, only the market and the capture rate change. Each row gives you the implied probability through mid price, the touch, the depth on each side, and the emit time you bracket kickoff on.
mid_priceImplied probability, 0-1- spreadTightens pre-game, can gap on big plays
bid_depth_totalask_depth_totalLiquidity that thins as risk spikesevent_timestampBracket kickoff and key moments on emit time
Studying the movement
- 1Find the sports markets for a fixture and pull their snapshots across a window that brackets kickoff and the full match.
- 2Align on event_timestamp so the pre-game and in-game segments sit on one clock and key moments line up across markets.
- 3Trace the implied probability through the three phases, drift, swings, resolution, and mark where spread and depth react.
- 4Aggregate across many fixtures of the same type so a pattern like pre-game drift separates from a single dramatic game.
~2 Hz is the book, not the scoreboard
Two limits worth stating plainly. The roughly 2 Hz rate is how often we sample the order book, fast enough for line moves and score steps, but not a frame-accurate record of the play itself. And settlement is Polymarket’s: we capture the book’s movement toward the outcome, while the final grading comes from the market’s own resolution, not from us reading the scoreboard.
What the swings can and can’t tell you
- A sharp in-game step usually marks a real game event being priced, but the snapshot shows the book’s reaction, not the play, pair it with the timeline to attribute it.
- Pre-game drift is most readable in aggregate; one game’s late move could be news, could be flow, and the data alone won’t always separate them.
- Depth thinning around big moments mirrors the defensive response seen elsewhere, makers pulling size when the next event is uncertain.
- Because sports samples slower than crypto, widen your time buckets so the arc stays smooth rather than blocky.
A sports market writes its whole story in the book, the patient drift before kickoff and the violent steps once the ball is live, and at ~2 Hz you can read it from first whistle to last.
Trace a game from the book
See sports coverage and the tiers it unlocks, or replay a fixture’s order book through kickoff and into play.
Frequently asked questions
Which sports markets are covered and at what rate?
NBA, NFL, EPL, and FIFA outcome markets, captured at roughly 2 Hz, slower than the ~20 Hz on crypto because sports books reprice less furiously most of the time. That rate is enough to trace pre-game line moves and catch the step changes when a score or momentum shift hits.
What tier do I need for sports data?
Sports sits on the Scale tier and above, alongside the other non-crypto categories and Hyperliquid exchange data. The free and Pro tiers focus on crypto; Scale unlocks sports, weather, economics, social, and equities.
Does the data tell me what happened on the field?
It tells you how the order book moved, not what the play was. A sharp in-game step usually marks a real game event being priced, but the snapshot is the book’s reaction, pair it with the match timeline to attribute a swing. Settlement itself comes from Polymarket’s resolution, not from us reading the scoreboard.
How do I study pre-game drift versus in-game swings?
Pull snapshots across a window bracketing kickoff and the full match, align on event_timestamp so pre-game and in-game segments share one clock, and trace the implied probability through drift, swings, and resolution. Aggregate across many fixtures so a systematic pattern stands out from one dramatic game.