Polymarket hockey data: 9k markets on the National League

Polymarket prices hockey as match winners, totals, and period markets. The live book is free to read from the official API, and it evaporates the moment the fixture settles.

That is the whole catch: the /book endpoint cannot answer 'what was this game priced at, period by period?' once the market has resolved. Period-level depth needs an archive that recorded it live — the PolyOrderbooks 250ms capture across the series it lists.

The catalog-wide 23,000 half-and-quarter-style markets include the hockey period markets, and each trades as its own book — a second-period moneyline settles independently of the game market.

The stored record is the same as every series: full bid/ask ladders with sizes, a captured_at timestamp, and the winning outcome on resolution, queryable at any point from opening to final horn.

• 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 hockey events; /v1/markets for questions and slugs.
• Coverage — measured counts per sport on the sports data page.

A National League game settles in three periods, and the second-period moneyline is its own binary long before the game market closes. The official API shows that ladder live; the archive records it every 250ms so a third-period tying goal can be replayed with the exact executable side at the second it landed.

Hockey trades match winners, totals, and period markets inside the 9,000-market family, with the quarter-style window shared with the catalog's 23,000 half-and-quarter markets — period binaries settle independently, so a single night yields several closed books.

The documented behaviors to verify: whether second-period lines are systematically wider after a first-period blowout, whether totals drift on power-play minutes, and whether the game-winner line tightens inside the final two minutes of regulation regardless of score.

The pull sequence is standard: /v1/series scopes the league and season, /v1/markets filters to the night's games, and the archive returns the full period window second by second.

The intermission boundaries mark the natural repricing points, and the final two minutes of a tie game contain the tightest quotes of the evening.

The 9,000-market family is small but real — fine for single-season period studies, not for sweeping claims.

Starter holds 3 days of 250ms (a full weekend), Pro 120 days, resolved fixtures queryable throughout.

Study one: test whether period markets trade wider after a first-period blowout than the game line, measured off the executable side with the one-sided share held constant.

Study two: replay a tying goal in the final two minutes of regulation and count the seconds until the game-winner line reprises; the venue's recorded sub-second moves are the benchmark.

Study three: compare power-play minutes against total-line drift to isolate whether the book reprises on man-advantage time or on shots.

A single season weekend inside the free 3-day window starts the picture; the Pro window to 120 days covers a full regular-season month and keeps resolved games queryable.

The starter exercise is one period: pull the second-period window of any National League game from the series list and count the executable-side flips around the intermission. Period binaries settle on their own clock, and the archive keeps that clock readable after the final horn.

Then compare the period book to the game winner at the same second and look for the lag — when the game line reprises off a period-market move, the record shows it frame by frame.

Resolve the pulled markets and keep the game queryable; one weekend of that routine is the smallest honest start for studying period structure across the family.

• Sports data
• Market data
• Order book data