Polymarket cycling data: race pricing at 250ms

Cycling sits in the sports long tail — families Polymarket lists continuously but reports without a counted split — alongside motorsport, MMA, basketball, cricket, and other event families.

The record is uniform: each cycling market PolyOrderbooks lists captures as a full L2 book four times per second, while the official API serves only the live, current-moment ladder.

Cycling race and stage questions resolve to the shared record — bid and ask ladders with sizes, a captured_at timestamp, and the winning outcome once settled.

No cycling-specific count is quoted because the family is measured as long tail; the honest indicative sizes live on the sports data page.

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

A Grand Tour stage runs hours on a flat line and then a summit finish reprices the winner inside one climb. The official API serves the live ladder; the archive keeps the final climb's frames so the breakaway's effect on the GC line can be replayed after the podium.

Cycling trades in the 700k+ long tail alongside motorsport, MMA, basketball, and cricket: stage and GC-winner questions, series listed continuously, counted in the combined sports pool rather than split. The unit of study is the stage clock — a breakaway can repurpose an entire day's book in one attacking kilometre.

The collectible measurements: whether GC-winner lines survive flat stages untouched, whether the summit-climb stage reprises both the stage and GC books inside the final two kilometres, and whether early breakaways are priced as noise until the gap crosses a threshold the book respects.

Practical access follows the shared chain: /v1/series for the tour and stage, /v1/markets for the questions, then the archive window across the decisive climb for the full ladder.

The final climb is the seam: the executable side flips on summit attacks in a way flat days never display.

As long-tail data, a single stage is a case study; a full tour in one archive window is the honest sample.

Starter covers 3 days of 250ms — several stages — and Pro reaches 120 for the whole three-week tour.

Study one: measure summit-climb repricing across a three-week tour, counting frames from each attacking kilometre to the new executable price on both stage and GC books.

Study two: compare flat-stage GC-winner books against climb-stage books to test whether the flat line is genuinely inert at 250ms.

Study three: test whether breakaway gaps are priced as noise until the gap crosses the book's threshold, locating the exact frame of repricing.

One tour stage on Starter starts the record; the full tour in Pro's 120-day window is the honest sample for GC conclusions, and the stage-level data stays queryable after the podium.

Replay a mid-tour sprint stage after a summit finish, and the contrast in executable-price behavior between the two archetypes makes a clean teaching frame for a new reader of the archive.

Start with one summit finish: pull the final-climb window of a tour stage and count the executable-price moves across the attacking kilometres. The stage and GC books reprice at slightly different moments, and the archive keeps both at 250ms.

Then pull a flat stage of the same tour and compare how inert the GC book genuinely is; the untouched flat line is the honest control the catalog needs.

Resolve the stage and keep it queryable; a few stages on Starter start the record, and the full tour in the Pro window is the smallest legitimate sample for GC conclusions.

• Sports data
• Market data
• Order book data