Kalshi KXMARATHON — Marathon Winner Data

A series-day bundle contains every Marathon Winner contract that was listed on Kalshi on that UTC day — traded or not — 73 contract files on the latest covered day. Each file carries the full raw feed: every trade print with price, size and side, plus Level-2 order-book updates, in the same zstd-compressed schema as our crypto venues. Event contracts quote 0..1; turnover is USD-normalized.

Coverage runs April 2026 – present — 81 trading days, 75.0 MB compressed in total. A day typically holds 77 contract files (up to 158 on the busiest day, 24.8 MB). Pick individual days in the shop calendar for €1 each; delivery is an instant per-file download or one archive for the whole day.

The 10 most traded Marathon Winner contracts on 2026-11-01 — 2 closed Yes (≥ 0.97), 7 closed No (≤ 0.03), 1 still open at the last print: 2,012 trades, $14.9K USD turnover. The busiest contract was KXMARATHON-NYC26-AALM with $9.2K turnover.

Prices quote 0..1 (probability of Yes). Open/high/low/close/VWAP are that UTC day’s trade prints per contract; the implied outcome reads the last print — ≥ 0.97 as Yes, ≤ 0.03 as No, anything between as still open — venues publish the official result. Buy share = taker-buy side of turnover; spread = the contract's turnover-weighted quoted bid-ask spread in cents, top-1 depth = mean best-bid/ask notional per side, L2 updates = order-book level updates recorded for the contract that day. “Settles” is the venue-listed close time of the contract where the venue publishes one (UTC).

On Kalshi, a Marathon Winner contract trades for about 74 days from open to close, is listed roughly 64 min before it opens, and quotes in 1 ¢ price steps (0..1). Measured on the 10 contracts captured on 2026-11-01.

KXMARATHON-NYC26-HESA traded in 5 of the day’s minutes (04:14–08:39 UTC): 17 trades, $303.89 turnover, opening at 0.010, ranging 0.010–0.060 and closing at 0.010.

Split by season and match day, never at random — the contracts of one slate share injuries, weather and the same public money, and a random split leaks the slate's results into training; walk forward over game days and purge the days whose contracts were still open. The sports-model guide works the closing-line benchmark, the pre-game line path and that split through.

Purchased data is for your own use — backtesting, research, analytics and training your own models, in-house or in your own products; the trained models and derived data you build from it are yours. The raw files and tick streams themselves may not be redistributed, resold or published. The full terms are on the data license page.

Agents: /llms.txt · MCP server /mcp · every page as markdown via Accept: text/markdown