What the market asked
The Polymarket event asked a narrow weather question: what would be the highest temperature in Madrid on August 27? Resolution depended on a single data source and rule set—the highest whole-°C value in the Temp column of the NOAA timeseries for LEMD (Adolfo Suárez Madrid-Barajas), with Weather Underground as backup if needed. Preliminary readings could shift until the following day’s first observation. The market was structured around discrete temperature outcomes rather than a simple binary, so the useful forecast was which specific high was most probable.
What the council concluded
On 2026-08-26 the council action was BUY_YES, with an overall probability of 22% and confidence of 58%. Only two models participated—Grok and GPT—with Grok as chairman. The graded basis was the council’s most-likely outcome, not a blanket yes/no on the full event.
The published reasoning stated: “Council recommends BUY_YES (YES). Probability: 22%, Confidence: 58%. Resolution uses the single highest whole-°C value in the Temp column of the NOAA timeseries for LEMD (Adolfo Suárez Madrid-Barajas) on 27 August 2026 (Weather Underground backup if needed); data are preliminary until the next day's first observation.”
Within that framing, the council’s most-likely specific outcome was “Will the highest temperature in Madrid be 23°C on August 27?” at 27%. That became the graded pick. The 22% council figure and the 27% most-likely figure are consistent with a multi-outcome temperature market: the models concentrated probability on 23°C as the mode while remaining cautious about any single bucket.
How the probabilities sat
A 27% call on the eventual winner is not a high-conviction stance. It implies the council saw a relatively flat distribution across nearby temperatures and simply ranked 23°C first. The accompanying 58% confidence and the explicit note about preliminary NOAA data show the models were aware of both forecast uncertainty and settlement mechanics. No stronger edge was claimed.
What actually happened
The winning outcome was exactly the graded pick: the highest temperature in Madrid on August 27 was 23°C. The council was correct under the “most_likely” grade basis. Because the assigned probability was only 27%, the Brier score landed at 0.5329—(0.27 − 1)²—which is the arithmetic consequence of a modest probability on a correct binary-style resolution of that bucket. Correct direction, limited sharpness.
Takeaway
This was a clean hit on a thin-edge weather market. The council identified the right temperature bin a day ahead, respected the official LEMD/NOAA resolution path, and did not overstate conviction. The mediocre Brier score is a reminder that “most likely” and “highly likely” are different claims; getting the mode right still leaves substantial residual uncertainty when neighboring outcomes remain plausible. For weather markets settled on a single station reading, that residual is often the story.
AI-generated analysis for informational purposes only. Not financial advice.