Studies

The same city, the same day, and one time in five the two venues disagree

On 657 matched temperature markets, Kalshi and Polymarket settled to opposite outcomes 19.5% of the time. Counted city by city, with what trading the gap would have cost.

The Heron

The Heron

— 2 min read

By The Heron. Figures from both venues' own settlement records, graded on 23 September 2026.

Kalshi and Polymarket both run daily markets on the high temperature in the same American cities. Same city, same date, same two degree band. You would expect them to settle the same way every time.

They do not. On 657 matched markets across five cities, the two venues settled to opposite outcomes 128 times, or 19.5% of the time. On 69 of the 169 city days we could match, at least one band flipped between the two venues.

This is not a rounding quirk on one side. The disagreements split almost exactly down the middle: Kalshi said yes where Polymarket said no 65 times, and the reverse 63 times. And they come in pairs. When Austin's 100 to 101 band settled yes on Kalshi and no on Polymarket on August 5, the 98 to 99 band did the opposite. The two venues read two different temperatures for the same city on the same day, and each one settled correctly against its own reading.

We wrote in August that the venues use different weather stations. This is what that costs, counted.

Chicago, every disagreement

Chicago had the most of any city we matched: 35 flipped bands out of 156 markets, on 20 of its 39 days.

Date Band Kalshi Polymarket
2026-07-25 80-81 no yes
2026-07-31 80-81 no yes
2026-07-31 82-83 yes no
2026-08-10 86-87 no yes
2026-08-10 88-89 yes no
2026-08-13 76-77 yes no
2026-08-13 78-79 no yes
2026-08-16 84-85 no yes
2026-08-16 86-87 yes no
2026-08-18 78-79 no yes
2026-08-18 80-81 yes no
2026-08-19 78-79 no yes
2026-08-19 80-81 yes no
2026-08-22 80-81 no yes
2026-08-22 82-83 yes no
2026-08-27 80-81 yes no
2026-08-27 82-83 no yes
2026-08-28 82-83 no yes
2026-08-28 84-85 yes no
2026-08-29 80-81 no yes
2026-08-29 82-83 yes no
2026-09-02 94-95 no yes
2026-09-02 96-97 yes no
2026-09-03 90-91 no yes
2026-09-03 92-93 yes no
2026-09-09 80-81 yes no
2026-09-11 80-81 yes no
2026-09-11 82-83 no yes
2026-09-12 86-87 no yes
2026-09-12 88-89 yes no
2026-09-14 72-73 yes no
2026-09-17 78-79 no yes
2026-09-18 70-71 no yes
2026-09-18 72-73 yes no
2026-09-19 72-73 yes no

Why this matters if you trade either venue

If you have ever looked at a temperature market on one venue and priced it off the other, you were pricing a different contract. A cross venue "arbitrage" in these markets is a bet that two thermometers agree, and one time in five they do not.

We tested that bet properly rather than assume, and the answer is below the line: the direction of the disagreement, which venue actually forecasts better when you sample both at the same instant, and what trading Kalshi against Polymarket's price would have returned.

Below the paywall: all five cities with counts and direction, every one of the 128 disagreements by date and band, calibration of both venues at three times of day on a matched clock, and the return from trading Kalshi on a 10 to 25 cent disagreement with Polymarket. Member is $15 a month, or free for seven days.

All five cities, the matched clock comparison, what trading Kalshi against Polymarket's price would have cost, every disagreement, and the method are in the member appendix.

Watch the 20 second version

More on the Pondletter channel

Read more