Question 1

How do weather betting markets work?

Accepted Answer

Kalshi and Polymarket both list daily contracts on a city's high temperature for a single day. Each contract covers one bucket, such as 85 to 86 degrees Fahrenheit or 90 degrees and above, and the market price reflects the probability traders assign to that bucket happening. Contracts typically open about a day and a half before the measurement day and close near midnight, then settle once the National Weather Service publishes its official climate report for that station the next morning.

Question 2

Which station settles each city?

Accepted Answer

Every city on this page settles against one specific National Weather Service reporting station, not the whole metro area. New York settles at Central Park, Chicago at Midway, and so on for every city we track. The venue ladder panel on each city's detail view names the exact settlement station for both Kalshi and Polymarket, since the two venues do not always pick the same one.

Question 3

Why do Kalshi and Polymarket disagree on NYC?

Accepted Answer

Kalshi settles New York against Central Park. Polymarket settles against LaGuardia. Central Park sits inland with tree cover that runs a degree or two cooler on hot afternoons, while LaGuardia sits on open pavement near the water and tends to run warmer. Since the two venues are reading different thermometers, their contracts can honestly price different outcomes on the same day without either one being wrong.

Question 4

How accurate are Kalshi weather markets?

Accepted Answer

The Record section on this page tracks it directly: every graded contract's price band against how often that band actually happened, with the sample size published next to every point. A well calibrated market lands close to the diagonal reference line, where a band priced around 70 cents hits close to 70 percent of the time. We publish this chart continuously rather than a single historical number, since accuracy varies by price band and by venue.

Question 5

How GhostLeg computes ensemble probabilities

Accepted Answer

For each city's settlement station, GhostLeg pulls two independent public forecast-model ensembles through Open-Meteo: the ECMWF ensemble (51 members) and the GFS ensemble (31 members), 82 members in total. Each member produces its own forecasted high for the climate day, and after correcting for that station's own historical bias, we count the share of members whose forecast lands in each contract's bucket. That share is the model probability shown on the board and on every city's ladder.

Question 6

The climate day rule (local standard time)

Accepted Answer

A city's climate day always runs midnight to midnight in that city's local standard time, not local clock time. During daylight saving time this means the measurement window actually runs 1 AM to 1 AM on the clock, because the official National Weather Service climate day never shifts with the clock change. Getting this boundary right matters: get it wrong and an hour of the wrong day's temperatures leaks into a contract's grading.