The two venues list the same contract — is BTC above where the 15-minute window opened — and differ in three things that change the answer: Kalshi settles on a 60-second average of CF Benchmarks' BTC Real-Time Index, Polymarket settles on a single Chainlink print at the boundary, and Polymarket resolves ties Up. Fees are now within half a basis point of each other.
Facts only. Sardine has not published settlement history for Polymarket and no figure on this page describes how Polymarket contracts have settled.
| Kalshi KXBTC15M | Polymarket BTC 15m | |
|---|---|---|
| Contract shape | One binary per window: is BTC above the window open? | One binary per window: is BTC above the window open? |
| Strikes per window | One. 0 of 6,995 settled windows carried more. | One. |
| Settlement index | CF Benchmarks BRTI | Chainlink BTC/USD stream |
| Settlement endpoint | 60-second average of roughly 60 index prints | A single print at the window boundary |
| Ties | Not applicable: an average almost never ties exactly | Resolve Up (the condition is ≥) |
| Window length | 15 minutes, opening on the quarter hour | 15 minutes, opening on the quarter hour |
| Window named by | Close time, in the ticker (KXBTC15M-YYMMMDDHHMM) | Start time, in the slug |
| Taker fee | 0.07 × C × p × (1 − p), rounded up to the cent per order | 0.0695 × C × p × (1 − p), banker's rounding (US, effective 2026-09-17) |
| Maker fee | None | −0.0125 × C × p × (1 − p), a rebate |
| Settlement history published by Sardine | Yes, see the base rates page | Not yet published |
An average of sixty prints is harder to move than one print. A contract settled on a single reading is exposed to whatever the index happened to say at that instant, including a spike that reverses immediately afterwards. The two rules therefore do not always agree on the same window, and the difference is largest where the price sat closest to the strike.
Sardine holds a reconstruction of the Polymarket rule applied to its own price history, and it is not good enough to publish. The reconstruction substitutes a third index for both venues' own, and that substitution moves the result by more than the difference being measured, so any percentage it produces would be an artefact of the proxy rather than a property of Polymarket. A recorder is now capturing Polymarket's own book and its own resolved windows; numbers will appear here when that data passes the same split-half calibration gate every other published number passes, and not before.
Sources on this page: kalshi_vs_polymarket_btc_15min.csv.
Cite as: Sardine (sardine.info), CC BY 4.0. Licence: CC BY 4.0.