How temperature markets settle, and why your weather app shows a different number

Temperature markets are the most concrete contracts on any exchange. A number is measured, compared with a threshold, and the contract pays. They are also where new traders lose money fastest, because the number that settles the contract is not the number on their phone.

Four numbers that are not the same

The forecast. What a model or a forecaster expects the high to be. It changes through the day and it settles nothing.

The city's weather. New York City has many observation sites. A high of 87 in one neighborhood and 84 at the airport are both "the weather in New York." The contract does not care.

The named station's reading. The rule names one station. As of September 2026, Kalshi's daily New York high-temperature contracts name station CLINYC and specify The Weather Company as the source; confirm the current rule on Kalshi's own market page before relying on it. The reading at that station, from that provider, is the only number that matters.

The preliminary versus the final report. Observation sources publish preliminary readings and later finalize them. A rule names which one settles. A preliminary reading one degree above a threshold can be revised below it.

Our temperature trackers show the rule text on every contract because the named station and source can differ by city and by contract, and they can change.

How the ladder reads

A daily high-temperature market is a ladder: 83° or below, 84° to 85°, 86° to 87°, and so on, up to 92° or above. Exactly one rung settles YES. The prices across the rungs are the market's probability distribution for the day's high, and the rung nearest 50 is roughly the market's median. As observations accumulate through the afternoon, the distribution collapses toward the actual reading, and by evening most rungs are at 1 and one is near 99. Then the source publishes, and one rung goes to 100.

The information that actually matters

For a daily high, the sequence of information is: the overnight forecast for the station's location, the morning observations at that station, the afternoon observations as the high is approached, and the source's published daily maximum. Each step reduces uncertainty. Note that a contract closing at midnight is often settled the next day when the source finalizes.

The National Weather Service publishes forecasts, alerts, and observations through a public API, and Kalshi's own guidance directs readers to check the named source on each market. That is the correct starting point for anyone who wants to understand these contracts: the structured source, not other publishers' summaries of it.

A worked example

The New York ladder on a September afternoon shows 83° or below at 59, 84° to 85° at 40, and everything above at 1 to 7. The market expects a high around 83 or 84 and is unsure which side of the line. A weather app shows 86 for "New York." The app is showing a forecast for a different location, or a different site's observation, and it is not evidence about the CLINYC reading. The contract settles the next day on the provider's daily maximum for that station.

Why we track and do not trade

Weather contracts settle daily, which makes them the best material for learning how these markets behave: every day produces a result, a prior price, and a lesson about how the distribution collapsed. We cover those settlements as news. We do not publish weather trade ideas, because claiming an edge in these markets requires station-level observation data and a model of how the source finalizes readings, and a claim without that is a guess dressed as analysis.

What to take away

• The contract settles on one station, one provider, one report. Read the rule.
• The forecast, the city, and the preliminary reading are not the settling number.
• The ladder is a distribution; the rung near 50 is the median.
• Ask for station-level evidence before believing anyone's weather edge.