Explainer · checked Oct 21, 2026
Can Kalshi's weather markets be gamed?
Three things can come between the weather and a market's result: a bad reading at the station, a reading you saw that doesn't count, and the source the rules name. Here is what the rules say about each, and what has happened, with the reports themselves.
The short answer
The market doesn't settle on a single sensor reading. It settles on the day's high as the climate report and The Weather Company publish it, after the NWS office's checks, and Kalshi can hold the result when that number is wrong. Bad readings have reached the hourly reports; in the cases below, they didn't reach the result.
A bad or tampered reading
The climate report is built from the station's own daily summary, which the NWS's guidance says is not checked by the local office before it's sent. The office checks the climate report: "WFOs should correct erroneous data in the CLI". That is where a reading the office judges bad is left out of the official high.
Source: NWS Instruction 10-1004, section 4.1.5
Central Park, Aug 27, 2026: an 80°F reading in heavy rain
Central Park's reports read 21.1°C at 3:42 PM EDT, 26.7°C (80°F) at 3:51 PM and 21.7°C at 3:59 PM, in heavy rain. The 6-hour and 24-hour maximum groups that followed said 27.2°C, which is 81°F.
The final climate report put the day's high at 77°F, at 2:50 PM. The market settled at 77, and the "79° or below" bracket won.
Source: Central Park's METARs that afternoon, via IEM; The final climate report (202608280657-KOKX-CDUS41-CLINYC); The settled markets, Kalshi's public API
Phoenix, Sep 16, 2026: a 24-hour maximum two degrees above everything else
The 24-hour maximum group on Phoenix's 11:51 PM MST report said 36.1°C (97°F). Its hourly readings and 6-hour maximums that day topped out at 35.0°C (95°F).
The final climate report said 95°F, and the market settled at 95: "95° to 96°" won.
Source: Phoenix's METARs that day, via IEM; The final climate report (202609170846-KPSR-CDUS45-CLIPHX); The settled markets, Kalshi's public API
When Almanac replayed calls of brackets already ruled out by the high so far (Aug 20 to Oct 18, 2026), 2 of the 7 failed, both because the NWS later discarded the station reading behind them: New York on Aug 27, 2026 and Phoenix on Sep 16, 2026, the two cases above.
What Kalshi's rules say about a wrong number
Every daily-high market from the September 3, 2026 events on carries this paragraph in its rules. The events up to September 2 don't; all 24 series changed on the same day.
“If it is determined in the sole discretion of the Exchange that there is a material error in the initial non-preliminary publication, Expiration may be held until the publication of a non-materially-erroneous data revision or until the Expiration Date, whichever comes first. If no data is available at the end of this period, all markets will resolve to the last fair price determined by Kalshi. Please note, a status of "Final" on weather.com/kalshi is provided for convenience only, and does not guarantee data to be error-free and thus exempt from the above.”
Source: Rules of KXHIGHNY-26SEP02 (without it), Kalshi's public API; Rules of KXHIGHNY-26SEP03 (with it); Kalshi contract terms (GLOBALTEMPERATURE)
A reading you saw that doesn't count
A METAR you watched isn't the settlement value, and neither is the afternoon preliminary report. From May 1 to Oct 20, 2026, across the 24 stations, the afternoon report's high matched the final report's on 3,890 of 4,348 station-days; it was higher than the final on 3. Kalshi's contract terms use only the first official non-preliminary report, and revisions after the expiration date don't count.
Miami, Aug 29, 2026: the NWS sent two final reports. The first, at 4:24 AM EDT, put the high at 90°F (3:25 PM); the second, at 5:10 AM, at 85°F (5:11 AM). The Weather Company and Kalshi both used 90, and the "90° to 91°" bracket won. Archives that keep the later report show 85, which is the one difference in the cross-check below.
Source: The first final report, 4:24 AM; The second, 5:10 AM; The settled markets, Kalshi's public API
How to read the climate report · Decode a METAR's exact reading
The switch to The Weather Company
From the Aug 14, 2026 events, the rules settle "according to The Weather Company" (weather.com/kalshi), then the NWS. From Aug 14 to Oct 18, 2026, Kalshi's value matched The Weather Company's on 1,541 of 1,541 settled days, and the NWS climate report's on 1,539 of 1,540.
Source: The Weather Company, weather.com/kalshi; Kalshi contract terms (GLOBALTEMPERATURE); Almanac's cross-check of 1,541 settled days
When a result waits
Of 1,800 events from Aug 14 to Oct 29, 2026, half settled within 6.2 hours of trading closing and nine in ten within 7.2 hours. Three took more than a day.
- San Diego, Oct 10, 2026: settled 7.4 days after trading closed.
- Oklahoma City, Sep 16, 2026: settled 6.4 days after trading closed.The Iowa Environmental Mesonet's archive holds no final climate report for that day, only two afternoon reports of 101°F. The market settled at 101.
- Louisville, Sep 6, 2026: settled 34 hours after trading closed.
Kalshi didn't say why. Results can also split: before these events, Miami's Apr 11, 2026 market settled on Apr 20 at $0.01 on "78° to 79°" and $0.99 on "80° to 81°", so a payout isn't always exactly $0 or $1.
Source: Close and settlement times of every event, Kalshi's public market data; Oklahoma City's climate reports for Sep 16, 2026, via IEM; Miami, Apr 11, 2026 (Kalshi's historical market data)
Questions
Can someone move the official high with a bad sensor reading?
A spike can show up in a station's hourly reports, but the market settles on the day's climate report, which the NWS office checks and corrects. On Aug 27, 2026 Central Park reported 80°F for one reading in heavy rain; the final report said 77°F and the market settled at 77.
What does Kalshi do if the published high is wrong?
From the Sep 3, 2026 events on, the rules let Kalshi hold expiration until a corrected value is published, up to the expiration date a week after the event; if none comes, every bracket resolves to the last fair price.
Did the switch to The Weather Company change the numbers?
Almost never. From Aug 14 to Oct 18, 2026, Kalshi's value matched The Weather Company's on 1,541 of 1,541 settled days and the NWS climate report's on 1,539 of 1,540.
Was the Paris hair-dryer case a Kalshi market?
No. News reports in April 2026 described abnormal readings at Météo-France's station at Paris-Charles de Gaulle airport on days a Polymarket Paris temperature market settled. Almanac couldn't verify the reports' details from a primary source.