Polymarket does not decide who wins its own markets. When a market closes, someone proposes an outcome to the UMA Optimistic Oracle and backs that proposal with a cash bond. A challenge window opens. If nobody disputes it, the proposal becomes the final answer and winning shares pay $1.00 each. If somebody does dispute it, the question escalates to a vote by UMA token holders, and that vote is final and on-chain.
That is the whole system in a paragraph. The rest of this article is the detail that actually costs people money: how long each stage takes, who is allowed to propose, what the four possible dispute verdicts are, and which contract wording tends to blow up.
This is an independent explainer, not official documentation and not legal advice. Bond sizes, timers, and eligibility rules are set by Polymarket and UMA governance and they do change — the official Polymarket resolution docs and the Polymarket Help Center are the authoritative sources, and this page links to them throughout. If you are new to the platform, start with What Is Polymarket? instead.
Who decides a Polymarket market’s outcome?
The UMA Optimistic Oracle does — a separate, independent protocol running on Polygon, not a Polymarket department. Polymarket is non-custodial and, for UMA-resolved markets, cannot reverse a finalized outcome by itself.
“Optimistic” describes the design philosophy: the system assumes a proposed answer is correct unless someone puts real money behind the claim that it is wrong. This is much cheaper than verifying every outcome from scratch, and it works because disputing a genuinely correct proposal loses you your bond.
Two practical consequences follow, and most resolution surprises trace back to one of them:
- The oracle grades the contract, not the news. It answers the question exactly as written, not the question you thought you were trading.
- There is no customer support appeal. Once the UMA vote finalizes on-chain, the result is immutable. There is no ticket you can open to undo it.
A small number of markets are resolved by Polymarket’s internal markets team rather than by UMA. When that is the case, it is stated in the market’s own resolution rules — which is one more reason to read them.
The resolution rules that matter are on the market page
Every market carries its own written criteria, found in the Rules section on the market page, usually beneath the order book. That text is the contract. It typically specifies:
- What counts as Yes and what counts as No, including thresholds and definitions.
- The source of truth — a named agency, league, official account, API, or publication.
- The deadline, including the time zone, which is a far more common source of disputes than people expect.
- Edge-case handling — cancellations, postponements, ties, revisions, or the event simply not happening.
If a market’s rules and a headline disagree, the rules win. “Everyone knows what happened” is not a resolution criterion.
The resolution lifecycle, step by step
There are three ways a market can travel from close to settlement.
Flow 1 — no dispute (the common case). A proposer submits the outcome with a bond. The challenge window passes quietly. The market resolves, the proposer gets their bond back plus a reward, and payouts execute.
Flow 2 — one dispute. Someone challenges the proposal by posting a matching bond. That voids the first proposal and opens a fresh proposal round. If the second proposal goes unchallenged, it is accepted and the market resolves.
Flow 3 — two disputes. The second proposal is also challenged. Now the question escalates to UMA’s Data Verification Mechanism (DVM), where UMA token holders vote to determine the outcome.
Note the structure: the first dispute does not trigger a vote. It triggers a do-over. Only a second dispute sends the question to the DVM.
How long does Polymarket resolution take?
Timings depend entirely on which flow the market falls into.
| Scenario | Typical elapsed time | What is happening |
|---|---|---|
| Undisputed | A few hours after the event | Proposal submitted, ~2-hour challenge window passes, market resolves |
| One dispute, resolved by re-proposal | Roughly half a day | First proposal voided, second proposal posted and accepted |
| Escalated to a DVM vote | Around 2 to 4 days | 24–48 hour debate period, then roughly 48 hours of token-holder voting |
| Repeated disputes or genuinely ambiguous wording | Up to a week or more | Multiple rounds, clarifications, contested evidence |
The challenge window is about two hours — short enough that if you intend to dispute something, you need to already be watching. Payouts themselves execute within minutes once resolution finalizes; the waiting is in the oracle process, not the settlement.
The practical takeaway for anyone holding size into resolution: the gap between “the event obviously happened” and “the money is in my balance” is a window in which your position can still change value. Treat it as part of the trade.
Who is allowed to propose an outcome?
This changed materially after 2025. Proposal rights used to be open to anyone willing to post a bond, which produced a steady stream of disputes from proposers who had not read the market’s specific quirks. UMA governance responded with a managed version of the Optimistic Oracle that restricts proposing to a whitelist.
As of writing, whitelist eligibility is based on a track record rather than an application: a minimum number of historical proposals over a rolling six-month window, at a high accuracy rate (roughly 95% or better), with addresses removed when they stop meeting the bar. There is a quirk worth noticing in that math — at the minimum proposal count, 95% accuracy effectively demands a perfect record, because one bad call out of five is 80%. Only proposers with a couple dozen submissions behind them have any margin for error.
The bond, as of writing, is typically $750 per proposal, and a disputer must post a matching amount. Both numbers are governance parameters and can be changed, so confirm current values on the UMA oracle dapp before acting on them.
For traders who never intend to propose anything, the relevant insight is this: proposals now come from a small pool of experienced, repeat participants whose bonds are at risk. That does not make them infallible, but it does mean a proposal appearing on your market is usually a considered call rather than a random guess.
What happens when a market is disputed?
Once a question reaches the DVM, there is a debate period of roughly 24 to 48 hours in which anyone can submit evidence — in practice this happens in UMA’s Discord, in the #evidence-rationale and #voting-discussion channels. UMA token holders then vote, which takes approximately another 48 hours.
The vote produces one of four verdicts, and the bond changes hands accordingly:
| Verdict | What it means | Bonds |
|---|---|---|
| Proposer wins | The original proposal stands | Proposer recovers their bond plus half the disputer’s bond; disputer forfeits theirs |
| Disputer wins | The proposal is rejected and a new one is needed | Disputer recovers their bond plus half the proposer’s; proposer forfeits theirs |
| Too Early | The underlying event has not actually concluded yet | Treated as a proposer error — disputer is rewarded, proposer forfeits |
| Unknown / 50-50 | No listed outcome fits; used rarely | Market settles at $0.50 per share on both sides; proposer forfeits |
Two of these deserve emphasis because traders routinely forget they exist.
“Too Early” is the verdict for proposals submitted before the event genuinely finished — a match still in progress, a deadline not yet passed. It is a proposer mistake, not a statement about the eventual outcome, and the market simply goes back into the queue.
“Unknown / 50-50” is the escape hatch for questions that turned out to be unanswerable as written. Every share redeems for $0.50 regardless of which side you were on. If you bought Yes at 8¢ this is a windfall; if you bought Yes at 92¢ believing you had a near-certainty, it is a serious loss on a position you were “right” about. Badly worded markets carry this tail risk and the price rarely reflects it.
Can Polymarket overturn a UMA resolution?
Generally, no — and you should trade as though the answer is no.
The nuance is that Polymarket has, on rare occasions, stepped in around an outcome it considered plainly wrong. In one 2024 case involving a market about Barron Trump’s involvement in a crypto project, the company overrode the UMA result and refunded traders, describing the oracle’s conclusion as incorrect.
But it has also declined to intervene in cases that traders found equally egregious. In March 2025, a roughly $7 million market on whether Ukraine would agree to a minerals deal with the Trump administration resolved Yes despite no deal having been signed, after a large UMA holder voted a substantial share of the total. Polymarket called the situation unprecedented and acknowledged the market had resolved too soon, but did not refund affected traders.
The pattern to internalize: intervention is discretionary, rare, and not something you can price into a position. Plan for the on-chain result to stand.
This is also why concentration in UMA’s token-weighted voting is a genuine, structural risk factor rather than a theoretical one. The proposer-side whitelist tightened who can start the process; it did not change who decides a contested vote.
Five rule patterns that cause disputes
Disputes cluster around recognizable wording, and you can screen for it before you enter.
Undefined time zones. “By December 31” resolves differently in UTC than in ET. If the deadline is close to the event, the time zone is part of your thesis.
Subjective descriptors. Anything requiring a human judgment call about whether something counts. The July 2025 market on whether Volodymyr Zelenskyy wore a suit — with roughly $200 million at stake — became the canonical example: mainstream outlets described the outfit as a suit, and the market still resolved No. The lesson is not that the voters were right or wrong. It is that “is this a suit” was never a question with a mechanical answer, and the contract had no tiebreaker.
Source hierarchies with no fallback. If the named source goes silent, retracts, or revises its figure, what happens? Good rules say. Bad rules leave it to a vote.
Revisable data. Economic statistics, vote counts, and official tallies get corrected after first publication. Does the market grade on the initial print or the revision?
Compound conditions. “X and Y by date Z” has more failure modes than it looks like, especially when X and Y resolve at different times.
If you cannot answer these questions from the rules text alone, assume the trader on the other side can — and price the ambiguity as a real cost, not an inconvenience.
Where to find the official rules and live status
Because this page is an explainer and not a source of truth, here is where to go for each thing you might actually need:
- Concept documentation — docs.polymarket.com/concepts/resolution for the current mechanism and parameters.
- Step-by-step help articles — the Polymarket Help Center covers proposing and disputing as user-facing procedures.
- Live oracle status — oracle.uma.xyz shows active proposals, challenge windows, bond states, and past requests.
- A specific market’s criteria — the Rules section on that market’s own page. Nothing else overrides it.
A pre-trade checklist for resolution risk
- Read the rules section end to end before sizing, not after something goes wrong.
- Confirm the time zone and the exact deadline, especially on markets that resolve near their cutoff.
- Identify the named source and ask what happens if it is late, silent, or revised.
- Look for a subjective word. If the outcome depends on someone’s interpretation, expect a dispute and size accordingly.
- Check whether the market is UMA-resolved or handled by Polymarket’s internal team.
- Decide in advance whether you will hold through the challenge window or exit before proposal.
- Set up notifications so a dispute does not reach you hours late — the Disputes and Timeline sections in Polymarket Alerts track challenged markets and their status, and the Features page shows how they work alongside price and whale alerts.
Frequently asked questions
How long is the Polymarket dispute window? About two hours from the moment an outcome is proposed. Miss it and you cannot challenge that proposal.
How much does it cost to dispute a Polymarket resolution? You post a bond matching the proposer’s, typically $750 as of writing. Lose the vote and you forfeit it, so this is not a low-stakes protest mechanism.
What happens to my shares if a market resolves 50-50? Every share, Yes and No alike, redeems for $0.50. This is the “Unknown” DVM verdict and it is used rarely.
Can a resolved Polymarket market be changed later? Once UMA’s DVM finalizes an outcome on-chain, it is immutable. Polymarket has occasionally issued refunds around outcomes it disagreed with, but this is discretionary and rare.
Does Polymarket use UMA for every market? Most markets, but not all. A subset is resolved by Polymarket’s internal markets team, and those markets say so in their rules.
Related reading
- How Polymarket Resolution Works — the shorter conceptual overview of the same process.
- Polymarket Resolution History: How to Look Up Past Outcomes — how to check how any market actually settled.
- Polymarket Trading Basics — execution, liquidity, and sizing around these events.
- Prediction Markets vs Sportsbooks — why a graded contract behaves differently from a posted line.
Bottom line
Resolution on Polymarket is a mechanical process with real deadlines, real money at stake for the people driving it, and no appeals desk at the end. The proposer posts a bond, the challenge window runs about two hours, a second dispute sends the question to a token-holder vote, and one of four verdicts settles it permanently. Everything that goes wrong tends to go wrong upstream of that machinery, in a sentence of contract wording nobody read carefully. Read the rules first, treat ambiguity as a priced risk, and know where to watch when a market you hold enters the queue.
This article is educational and independent. Bond amounts, timers, and eligibility rules are governance parameters that change — verify current details in the official Polymarket and UMA documentation before trading.