What the market asked
The Polymarket event was simple on the surface and strict in the fine print: would Bitcoin be above a stated threshold on August 26? The graded line was whether BTC would be above $64,000 on that date. Resolution did not hinge on a daily close, a high, or a vague “around then” feel. It depended on a single reference—the Binance BTC/USDT 1-minute candle close at 12:00 Eastern Time on 26 August 2026.
That design makes the contract a pure snapshot. Path noise, liquidity, and order flow in that exact minute can decide the outcome even when the broader day looks calm.
What the council concluded
Report date was 2026-08-25. Only two models participated—GPT and Grok—with Grok as chairman. The council’s action was BUY_YES, with a stated probability of 44% and confidence of 58%.
Their reasoning, quoted in full from the report:
“Council recommends BUY_YES (YES). Probability: 44%, Confidence: 58%. Resolution depends strictly on the Binance BTC/USDT 1-minute candle close at 12:00 Eastern Time on 26 August 2026. It is a single-minute snapshot, not a daily close or intraday high, so path noise and exact order-flow at that minute matter.”
In plain terms: they knew the contract was brittle to microstructure, still preferred YES, and did not claim high conviction. A 44% probability with 58% confidence is a modest lean, not a strong directional bet.
What the market itself was pricing
The council’s own probability mark was 44% on YES. That is the figure attached to the graded pick. In other words, the desk was not treating a move above $64,000 as the base case; it was treating it as a meaningful but minority chance—and still chose to buy YES against that view. The action and the probability sit in mild tension: BUY_YES at well under even odds implies they saw YES as cheap relative to their edge, even while assigning it less than a coin-flip chance of resolving true.
No separate external mid-market quote is part of this graded record beyond that council mark. What matters for the grade is the action they took and the probability they published.
What actually happened
The winning outcome was NO. Bitcoin was not above $64,000 on the resolution snapshot. The council’s graded pick was therefore incorrect. council_was_correct is false; grade basis was the action (BUY_YES). The Brier score on the call was 0.1936, which matches a 0.44 forecast on an event that resolved to 0: $(0.44 - 0)^2 = 0.1936$.
So the miss is not mysterious. They bought the side that lost, at a probability that was not extreme, and the binary went the other way.
Short takeaway
Two lessons stand out. First, honesty over narrative: the council was wrong, and publishing the loss is the point of the exercise. Second, snapshot crypto markets punish overconfidence in level calls. The council’s own write-up flagged path noise and one-minute order flow; that caveat was right, and the directional lean was not. A 44% YES with only medium confidence is the kind of edge that can be real in aggregate and still lose on any single print. When resolution is a single candle, “mostly right about the range” is not the same as “right about the contract.”
AI-generated analysis for informational purposes only. Not financial advice.