The Market Question
The Polymarket event asked a straightforward weather question: what would be the highest temperature recorded in Los Angeles on June 5? The market was structured as a set of narrow temperature bins, each framed as a yes/no contract (for example, “Will the highest temperature in Los Angeles be between 56-57°F on June 5?”). The council issued its report on 2026-06-04, one day before the observation date.
What the Council Concluded
Three models participated—Grok, Claude, and GPT—with Grok acting as chairman. The council’s graded pick, selected on the “most_likely” basis, was the 56-57°F bin. It assigned that outcome a 100% probability. Overall council action was recorded as BUY_NO at a 5% probability with 75% confidence.
The published reasoning stated:
“Council recommends BUY_NO (NO). Probability: 5%, Confidence: 75%. Latest NWS and coastal forecasts confirm a 68-71°F center for KLAX on June 5 with marine layer suppression; historical June patterns and recent actuals (69-72°F) reinforce this. The 64-65°F outcome wo”
The reasoning fragment itself points toward a mid-to-upper 60s / low-70s range and cites marine-layer suppression plus recent observations. Yet the formal most-likely designation and graded pick landed on the much colder 56-57°F bucket at full confidence. That internal mismatch between the narrative forecast and the selected contract is visible in the report data.
Market Context and Pricing
The supplied report does not include the contemporaneous Polymarket prices or order-book depth for the individual temperature bins. What is recorded is only the council’s own probability distribution: 100% on 56-57°F and a 5% probability attached to the BUY_NO action. No external market consensus figures are available in the graded record, so comparison is limited to the council’s stated view versus the eventual resolution.
What Actually Happened
The winning outcome was “Will the highest temperature in Los Angeles be between 70-71°F on June 5?” The council’s pick of 56-57°F was therefore incorrect. Because the graded probability on the chosen bin was 100% and the bin did not occur, the Brier score is 1.0—the maximum possible penalty for a binary probabilistic forecast.
Takeaway
This is a clear miss. The council’s own written reasoning referenced NWS guidance centered on 68-71°F and recent highs of 69-72°F, yet the formal most-likely selection and 100% probability were attached to a 56-57°F outcome roughly 15 degrees colder. Whether the discrepancy arose from a mapping error between forecast discussion and contract selection, an over-weighting of an outlier historical pattern, or simple mis-calibration, the result was a fully confident prediction that failed. Weather markets reward precise alignment between meteorological evidence and the exact bin being priced; here that alignment broke down. The public grade stands as a reminder that even when source forecasts are directionally sensible, the final probability assignment and contract choice must match them.
AI-generated analysis for informational purposes only. Not financial advice.