Polymarket soccer data: 325k markets, captured four times a second

Polymarket lists soccer questions league-wide — match winner, over/under lines, exact score, corners and cards — and its book endpoints serve the live ladder for any of them, free and without a key.

What those endpoints do not serve is history: the /book route returns the current moment, and depth that is not recorded while a market lives is gone. The PolyOrderbooks archive records that depth for every soccer market it lists — a full bid/ask ladder, four times per second.

Across the 700k+ sports catalog, over/under lines dominate at 372,000 markets, with player and team props at 58,000, corners and cards at 57,000, and exact score at 56,000. Soccer titles the same shapes: a match-winner question like 'Will RB Leipzig win on 2026-10-10?' sits beside total-goals Over/Under and exact-score lines.

Each of those markets resolves to the same archive record: full bid and ask ladders with price levels and sizes, a captured_at timestamp, and the winning outcome on resolved rows.

• Live ladder — the official /book endpoint, 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 and /v1/tags list soccer series; /v1/markets returns markets and slugs.
• Coverage — measured counts per sport live on the sports data page.

Walk a Premier League Saturday: the match-winner market opens days ahead, a second-half goal flips the favorite, and the final whistle settles the book. Each of those movements is a moment the live /book endpoint serves once and never again — the archive exists to keep the second-half second, the corner count, and the moneyline reset as queryable rows.

Soccer's settlement pace divides cleanly into halves, which makes it a natural study subject: the flurry between 75 minutes and full time in a 0-0 game is the tradeable window, and the stored record shows whether the ladder respected the draw or chased the last-minute lead.

Four angles have real literature behind them on this venue: whether favorites drift toward the bound as kickoff nears, whether the 2.5-goal line reprices on the hour, whether red cards reprice winners immediately, and whether stoppage-time goals were already embriefed into the book before the whistle.

For the practical pull, the /v1/series endpoint lists the event series by league — Premier League, La Liga, Serie A, Liga MX, FA Cup, MLS, and internationals — and /v1/markets filters the questions within each; the 250ms books then come off the archive REST in the same slug space.

Watch the interval boundary: the half-time mark moves the ladder less than the 75-to-90 window, and the minutes after the 60th are where lean favorites get priced at their tightest.

Soccer has the deepest recorded depth of any sport here (325k+ markets), so any study that needs a broad sample can reach it without touching the long tail.

The honest floor, as everywhere: 3 days of 250ms history on the free Starter, and paid windows to 120 days when the study crosses a fixture weekend.

Study one: measure whether winners drift toward the round bound during stoppage time of a 0-0 game, against the baseline that 98.6% of executable prices on the venue sit on the 0.99/0.995 grid. The half-time and 75-minute markers give the window its boundaries.

Study two: replay a midweek cup clash against a weekend league fixture from the same week and compare the one-sided share — the venue's nine-in-ten one-sided baseline — across both calendars.

Study three: pull the final five minutes of ten resolved league favorites and count how many settled at or inside one band of the closing bound; the executed corpus standard (298 resolved BTC contracts) is the analogue.

The honest minimum sample is one competition round in a single window — 60 to 90 fixtures at 250ms; anything narrower is a case study. The free 3-day Starter is exactly that size.

• Sports data
• Market data
• Order book data