How closely do Kalshi and Polymarket agree?

Both venues list contracts on the same games, so the same outcome carries two prices at once. We record every matched pair every five minutes. Below is what 309,533 paired observations across 379 games say about how those two prices behave. Every number here is computed from that history when the page loads.

They agree closely — and agree more as the game approaches

The disagreement between the two venues grows steadily the further from a game you look. On game day the two prices sit within a point of each other; three days out the typical difference is several times that.

WhenTypical gapOver 3 ptsSamples
Game day0.74 pts0.0%83,222
1 day before1.04 pts3.6%90,055
2 days before1.90 pts19.6%86,602
3 days before3.54 pts51.3%6,025
8 days before1.74 pts0.0%487
9 days before1.84 pts0.0%576
10 days before1.44 pts0.0%576
11 days before0.93 pts0.0%3,144
12 days before0.94 pts0.7%4,212
13 days before0.93 pts0.1%4,032
14 days before0.88 pts0.0%4,032
15 days before1.04 pts3.8%3,984
16 days before1.41 pts6.1%557
17 days before1.76 pts0.0%288
18 days before1.48 pts8.2%3,074
19 days before1.20 pts4.6%4,500
20 days before1.15 pts2.6%4,320
21 days before1.21 pts6.3%4,320
22 days before1.30 pts2.0%4,054
23 days before1.89 pts3.3%269

A gap is the difference between the two listed YES prices, in percentage points.

Most of the time, nothing moves

Between one five-minute observation and the next, the gap is unchanged 92.2% of the time, and the average move is 0.07 points. These are not fast-moving markets in the way an equities order book is. Checking them every few seconds would mostly re-read the same number.

Is the gap worth acting on?

Capturing a gap means buying on one venue and selling on the other, then paying Kalshi's fee. Across 137,801 readings of 178 games, the average edge after crossing both books is -1.63 points. It clears the books 6.2% of the time, and clears the fee as well 0.0% of the time. The second figure is the one that matters: an edge of a point is still a loss once roughly two points of fee land on it.

Those are not 137,801 independent trades: they are readings five minutes apart, roughly 774 per game, over a short window and mostly baseball. Treat it as a strong hint about these venues in this period, not a law. The figure updates as more games accumulate, whichever way it moves.

What that means if you trade it

The uncomfortable part: the gap is widest exactly where it is hardest to act on. Far-out games are thinly traded, and a wide bid/ask spread can easily exceed the difference between the two venues — so a listed gap is not the same thing as a profit. As a game approaches and liquidity improves, the two prices converge.

We would rather say that plainly than sell a screener that treats every far-out gap as an opportunity. A gap is a place to look, not a trade. The guide on price differences works through why they diverge, and the fee guide covers what Kalshi takes out of a winning contract.

Method and caveats

  • Every matched pair is sampled every five minutes and stored. The window here runs Aug 8, 2026 to Sep 2, 2026.
  • Pairs are matched deterministically on team codes and side, never fuzzily, and ambiguous cases such as doubleheaders are dropped rather than guessed.
  • Pre-game only. Once a game starts, one venue can settle while the other freezes, which manufactures spreads that were never real. Nothing here describes in-play behaviour.
  • This is mostly baseball. Football in season may behave differently, and the figures above will move as that history accumulates — they are recomputed from the data, not written down.
  • Horizons with fewer than 200 observations are omitted rather than reported with a number too thin to mean anything.

Polykal watches both venues around the clock and tells you when a gap opens that is unusual for that point in the schedule — because as the table above shows, a four-point gap means very different things on game day and three days out. See the live prices.