Polymarket’s resolution of two Strategy bitcoin-sale contracts has opened a sharper debate over prediction-market integrity. The platform resolved the May 31 market to “No” and the June 30 market to “Yes,” even though Strategy’s own filing said the company sold 32 BTC between May 26 and May 31.

The conflict comes down to timing. Strategy’s June 1 8-K disclosed that it sold 32 BTC during the May 26 to May 31 period for $2.5 million at an average sale price of $77,135. The economic event occurred before the May deadline, but public confirmation arrived one day later.

Confirmation Rules Override Event Timing

The market rules stated that the contract would resolve “Yes” if MicroStrategy sold any bitcoin by the date specified in the title, with information from MSTR, on-chain data and credible reporting listed as resolution sources. That wording left traders arguing that the sale date, not the filing date, should determine the outcome.

The resolution criteria relied on when they *sold* not when it was announced. Strategy's SEC-filed Form 8k explicitly stated that Strategy sold between May 26–31. A plain reading of the resolution criteria would suggest that the market should have resolved to YES, hence the…

— Galaxy Research (@glxyresearch) June 3, 2026

Polymarket’s final outcome created the opposite result. The May 31 contract resolved “No,” while the June 30 contract resolved “Yes,” turning the June 1 filing date into the effective settlement trigger. That made the market less about whether Strategy sold bitcoin and more about when the platform considered the sale confirmable.

Galaxy Research criticized the outcome, writing that “Everyone who bought YES predicted the future correctly,” and arguing that a prediction market loses value when resolution diverges from the event it was meant to price. The critique is severe because it targets settlement integrity, not only one disputed payout.

UMA Voting Faces Fresh Scrutiny

The dispute also revived concerns about Polymarket’s oracle architecture. Polymarket documentation says markets use UMA’s Optimistic Oracle, where anyone can propose an outcome, anyone can dispute it, and unresolved disputes can escalate to UMA token-holder voting. That structure makes final settlement dependent on a governance process as well as the written market rules.

Polymarket’s own help center says UMA token holders vote after a debate period and that finalized outcomes are immutable because Polymarket is non-custodial. That immutability protects settlement finality, but it also raises the stakes when traders believe the wrong interpretive standard was applied.

The operational risk is clear. Traders who modeled the contract as occurrence-based saw the May sale as sufficient. Traders and voters who applied a confirmation-based standard treated the June 1 filing as outside the May window. The same real-world fact produced different payouts because the evidentiary cutoff was ambiguous.

Event-contract exposure cannot be assessed only by the likelihood of the underlying event. Counterparties must also model platform-specific resolution language, oracle mechanics, dispute incentives and confirmation standards before sizing positions.

The governance lesson is equally direct. Prediction markets that settle high-volume contracts need rules that identify whether an outcome is event-based or confirmation-based at listing, not after money is committed. The Strategy dispute shows how retroactive interpretation can become the dominant market risk.

For Polymarket, the episode is a reputational test as much as a technical one. If prediction markets are expected to function as price-discovery tools for institutions, settlement must be deterministic, transparent and auditable. When traders predict the event correctly but lose on process, the product starts pricing adjudication risk alongside reality.