What the market asked
The Polymarket event asked a narrow weather question: what would be the highest temperature in Guangzhou on 26 August 2026? Resolution depended on a single data source—the highest whole-degree Celsius value in the Temp column of the NOAA/NWS WRH Time Series Viewer for station ZGGG (Guangzhou Baiyun International Airport) on the local calendar day. In practice this produced a set of related outcome markets, each tied to a specific integer temperature.
What the council concluded
Three models participated (Gemini, Grok, GPT), with Grok acting as chairman. The council’s formal action was BUY_NO at an implied probability of 21% and 68% confidence. Its internal ranking, however, identified one temperature as most likely:
Council recommends BUY_NO (NO). Probability: 21%, Confidence: 68%. Resolution is strictly the single highest whole-degree Celsius value appearing in the Temp column of the NOAA/NWS WRH Time Series Viewer for station ZGGG (Guangzhou Baiyun International Airport) on the local calendar day 26 August 2026.
Separately, the council assigned its highest individual probability—30%—to the specific outcome “Will the highest temperature in Guangzhou be 31°C on August 26?” That 31°C contract became the graded pick under the “most_likely” basis.
How the probabilities sat
The council did not treat any single temperature as a high-probability lock. Thirty percent on 31°C left the bulk of the probability mass spread across neighboring values, consistent with typical late-August uncertainty around a subtropical airport station. The broader BUY_NO stance at 21% reflected skepticism that any one aggressively priced alternative would clear the official ZGGG reading. No external market mid-prices are recorded in the graded report, so the only calibrated figures available are the council’s own 21% (action) and 30% (most-likely) estimates.
What actually happened
When the day closed and the ZGGG time series was checked, the highest whole-degree temperature was 31°C. That matched both the winning market outcome and the council’s graded pick. Because the grade basis was “most_likely,” the council was marked correct. The Brier score on the 30% forecast that verified was 0.49—exactly (1 − 0.30)²—showing a right-direction call that was still only modestly sharp.
Takeaway
A correct most-likely identification at 30% is useful but not especially decisive; weather markets at daily resolution remain noisy even when the data source is tightly specified. The council’s value here lay in correctly ranking 31°C at the top of a dispersed distribution and in anchoring resolution strictly to the stated NOAA/NWS station rule. Future similar events will test whether the same models can tighten probability mass without sacrificing the ranking accuracy shown on this Guangzhou print.
AI-generated analysis for informational purposes only. Not financial advice.