How UMA Decides Polymarket Disputes — and Why Traders Fight Them

On the international Polymarket exchange, the platform does not decide who won a prediction market. UMA's optimistic oracle does. Anyone can propose a result by posting a bond; if nobody disputes it within two hours, the market resolves. If the proposal is challenged, the question can escalate to a vote of UMA token holders.

That design is fast and cheap most of the time. It is also the source of Polymarket's most consequential resolution fights. When the wording of a market is ambiguous, or when a large token holder has a position in the outcome, the oracle becomes the market.

The mechanism, step by step

Polymarket's documentation sets out the process.

1. A proposer selects the winning outcome and posts a bond, typically $750, to UMA.
2. A two-hour challenge period begins. If no one disputes the proposal, the market resolves.
3. If someone disputes it, a second proposal round begins. The disputer posts a matching counter-bond.
4. If the second proposal is also disputed, the question goes to UMA's Data Verification Mechanism, where token holders vote.
5. Before the vote, there is a 24-48 hour debate period in which evidence can be submitted. The vote itself takes about 48 hours.

An undisputed market resolves in roughly two hours. A disputed one can take four to six days. During that time, the outcome is unresolved and payouts are not final.

The possible outcomes are not just "yes" and "no." UMA can also resolve "too early," which sends the market back for a later proposal, or "unknown/50-50," which pays each token 50 cents. Polymarket can also publish an "Additional context" clarification after trading begins. The documentation says a clarification cannot change the fundamental intent of the question, but it can tell proposers and voters which facts to weigh.

Case one: the $7 million Ukraine minerals market

In March 2025, Polymarket listed a market on whether Ukraine would agree to a minerals deal with the United States before April. On March 24 and 25, the "yes" probability jumped from about 9% to 100%, even though no official agreement had been announced.

The market resolved "yes." CoinDesk reported that the result drew suspicions of manipulation by a large UMA token holder, and that Polymarket responded in its Discord server by admitting the resolution was unexpected. Polymarket denied that the episode was a "market failure" that warranted refunds. It said the market had been resolved too soon, but defended the UMA voting process as the mechanism for correcting the outcome.

That distinction is important. Polymarket did not say the result was correct. It said the process was legitimate. For traders, those are different claims. A market can resolve through the rules and still produce an outcome that most observers regard as wrong.

The episode also exposed a structural asymmetry. Polymarket's markets can be very large, and a large UMA holder can have outsized influence over a vote. A holder who can move a meaningful share of the vote has a potential profit opportunity if they also hold a position in the market being resolved. That does not prove manipulation in any particular case. It does explain why traders watch the voter wallets as closely as the news.

Case two: the $160 million Zelensky suit market

In July 2025, Polymarket ran a market on whether Ukrainian President Volodymyr Zelenskyy would wear a suit before July. More than $160 million was at stake, according to CoinDesk.

Zelenskyy appeared at the NATO summit on June 24 in what many outlets described as a suit. The initial resolution was "yes." That proposal was disputed, and the "yes" price collapsed from about $0.19 to $0.04 as traders reassessed. The market ultimately resolved "no," according to Cryptonews, after the dispute process.

The fight was not really about whether the NATO outfit was a suit. It was about what the market's rules meant by "wear a suit," whether that outfit qualified, and which sources were authoritative. Those are exactly the questions a resolution oracle is supposed to answer. When the answer depends on interpretation, the oracle's vote becomes a judgment call rather than a data lookup.

Why ambiguity matters

Prediction-market titles are short. Resolution rules are longer, and the rules are what count. Polymarket's own documentation warns traders to read the rules before trading because "the market title describes the question, but the rules define how it resolves."

Ambiguity creeps in through several doors:

• Undefined terms. "Suit," "meet," "deal," "invade" and "out of office" can all be defined in ways that surprise traders.
• Timing. "By July" can mean before July 1, during July, or by the end of July. A single day can flip a market.
• Source selection. A market may resolve on a named outlet, an official announcement, or a consensus of reporting. If the sources disagree, the rules decide which one wins.
• Edge cases. What happens if an event is cancelled, postponed, or occurs in a form nobody contemplated? The rules may say "other" or may be silent.

When the rules are silent, UMA voters have to infer intent. That is where disputes become fights about fairness rather than fact.

What it means for traders

The first lesson is mechanical: read the full rules, not just the title. If the rules do not define the key term, assume the resolution risk is higher and size the position accordingly.

The second is that a disputed market is not a free option. The bond is real, the delay is real, and the outcome can depend on a vote you do not control. A trader who buys the "obvious" side after the news breaks may still lose if the rules point the other way.

The third is that the US venue is different. Polymarket US says the exchange itself determines outcomes using the criteria and sources in each market's rules, and that resolutions are final under its rulebook. The UMA process described here applies to the international exchange. A US trader who assumes the two venues resolve identically is taking a risk they may not have priced.

The broader point is not that UMA is broken. The bonded-proposal design is efficient when the outcome is clear. The point is that "the market resolved" is not the same as "the market resolved correctly." When the underlying question is ambiguous, the resolution mechanism is part of the bet.