Polymarket NFL data: 73k markets from kickoff to Super Bowl

Polymarket prices NFL and college football as moneylines, totals, and heavy half/quarter markets — a market for the second quarter of a Week 6 NFC game closes and settles like any other binary.

Those are live questions. For history, the official API ends at the current book: no end-of-quarter archive, no 'what was priced at halftime' lookup. Reconstructing football pricing after the fact requires an archive that recorded the depth as quarters elapsed — the thing PolyOrderbooks does at 250ms.

The catalog's 23,000 half-and-quarter markets are a football signature: 'Northern Illinois vs. Georgia State: 2Q Moneyline' is a real question Polymarket lists, and it trades as its own book.

Each game, each quarter, each prop stores the same way — full bid/ask ladders with sizes, a captured_at timestamp, and the winning side on resolution — so the second-quarter book is queryable eleven weeks later at the exact 250ms grid it was captured on.

• Live ladder — official /book, free and keyless, current moment only (what the API serves).
• Historical depth — full L2 books at 250ms over REST, free Starter with 3 days, paid windows to 120 days.
• Discovery — /v1/series for NFL and NCAA series; /v1/markets for questions and slugs.
• Coverage — measured counts per sport on the sports data page.

Sunday afternoon: the Chiefs moneyline opens modest, the first-half winner drifts after a slow opening drive, and a two-minute-drill touchdown reprises the full line inside one series. None of those past frames survive the final whistle on the official API; the archive replays the fourth quarter, second by second.

The catalog structures NFL and college football the same way: moneylines, totals, heavy quarter markets, conference championships, and season props, with each quarter settling like its own binary and each conference game keeping its own book.

Three patterns show up in the recorded history: quarter markets pricing tighter than game moneylines, red-zone series repricing winners within seconds of a first down, and garbage-time totals drifting late when benches empty.

The practical query is the Week-and-series pair — /v1/series lists the league and week; /v1/markets the game questions; and the archive returns the full 250ms ladder for any quarter window, from kickoff to kneel.

Watch the quarter boundaries and the two-minute drill at the end of each half — the ladder sees its widest repricing inside those minutes, not in the even flow between downs.

The 73k-market family keeps college and pro separated by series, so a study can stay on one side of the split without extra filtering.

Free Starter is 3 days; the whole week of games fits, and Pro widens to 120 days for season-long work.

Study one: measure whether quarter markets trade tighter than the game moneyline both pre-game and live, splitting by one-sided share so the executable side drives the comparison.

Study two: replay a two-minute-drill touchdown that reprises the total and count the second-level path from snap to score, calibrated against the recorded reprice behaviour in other families.

Study three: compare red-zone series repricing against first-down momentum — whether a first-and-goal tightens the winner marker faster than a first-and-ten conversion does.

A full week of games fits the 3-day Starter window at 250ms; the most recent comparable published size is the 298-contract resolved corpus, which sets the scale for what counts as evidence.

The practical first exercise is a single quarter: pull the second-quarter window of any NFL game from the current series list and count the executable-side flips inside the two-minute drill. Sub-second reprice timing is the documented norm, so the drill is where the quarter book earns its keep.

Then compare the quarter book to the game moneyline at the same second and note which one trades tighter; that single comparison introduces the multi-book structure of the entire catalog.

Finish by resolving the pulled markets and keeping the game queryable, so the Sunday becomes the seed of a per-week dataset instead of a one-off screenshot.

• Sports data
• Market data
• Order book data