How Are Polymarket Markets Resolved?

Polymarket markets settle through UMA's optimistic oracle: a proposer submits an outcome with a bond, a challenge window runs, and if nobody disputes, that outcome settles the market. Disputes escalate to a vote of UMA token holders. The market's rules text, written before trading starts, defines what counts as yes, what counts as no, and which sources decide. Everything downstream, including the rare wrong resolution, traces back to how clean or ambiguous that text is.

The pipeline, step by step

1. Rules text. Every market ships with resolution criteria: the exact question, the deadline, the resolution sources (an official result, a government publication, a named data source), and edge-case handling. This text is the contract; traders price against it, and disputes are argued from it.
2. Outcome proposal. Once the outcome is knowable, a proposer submits it to UMA's optimistic oracle and posts a bond. The oracle is “optimistic” because the proposal is presumed correct unless challenged; no vote happens in the default path.
3. Challenge window. A fixed window follows in which anyone can dispute the proposal by posting a matching bond. Most markets pass through undisputed, and the proposed outcome becomes final at the end of the window.
4. Dispute vote. If disputed, the question escalates to UMA's dispute-voting mechanism, where UMA token holders vote on the correct outcome under the rules text. The majority outcome settles the market, and the losing side of the dispute forfeits its bond.
5. Payout. After settlement, winning shares redeem at $1 and losing shares at $0.

Scheduled close is not settlement

A detail that regularly surprises traders: the end date on a market listing is the scheduled close of trading, not the moment your position pays out. Settlement arrives only after a proposal survives its challenge window, and a dispute adds a voting cycle on top. For markets whose rules wait on slow confirmation, an official certification, a signed agreement, a document release, the gap between scheduled close and settlement can stretch to weeks or longer. Capital in an unresolved market is locked the entire time, which is a real cost that quoted odds do not show.

Where resolution goes wrong

The oracle is an economic mechanism with bonds and votes, not a truth machine, and its output is only as good as the rules text plus the voters reading it. The known failure modes: ambiguous wording that two honest readers resolve differently; resolution sources that move slower than the market expected; and disputed proposals where the token-holder vote lands on a reading the underlying facts do not support. In one widely reported March 2025 case, a market on a Ukraine minerals deal resolved yes although no such agreement had been concluded, and Polymarket acknowledged the outcome did not match the underlying event. Such cases are rare against the volume of cleanly settled markets, but they set the tail risk for anyone treating a resolved price as ground truth.

How Convexly reads resolution quality

Resolution reliability is one pillar of the Market Trust card, Convexly's market-level diligence surface. Stated honestly: that pillar is currently a heuristic, built from subjective-wording checks on the rules text, a liquidity floor, time-to-resolution proximity, and category priors, and its thresholds are not yet calibrated against UMA dispute history. The card as a whole is a canary-preview research contract; it is not investment advice, not a compliance certification, and not a claim that a market will resolve correctly. The published methodology page carries the weights, gates, and the current promotion blockers, and the live cards are browsable at convexly.app/market-trust. Until an outcome ledger calibrates those heuristics, the honest use of the card is as a structured checklist with receipts, not a verdict.

The practical takeaway for a trader: read the rules text before the odds, check whether the resolution source can actually publish by the deadline, and treat any market whose wording you cannot paraphrase in one sentence as carrying resolution risk that the price does not display.