How a prediction market resolves

A market titled “Will the central bank cut rates by June?” is not a bet on that sentence. It is a bet on a paragraph underneath it that says which committee, which announcement, what counts as a cut, what happens to an emergency meeting, and which published figure is authoritative if the announcement is ambiguous. That paragraph is the instrument. The title is a search term.

This is why two markets on apparently the same event can trade a long way apart and both be correctly priced. If one settles on the date a decision is announced and the other on the date it takes effect, a decision announced in one month and effective the next resolves them differently. Nothing is wrong with either market; they are different contracts.

The practical consequence is unglamorous. Before trading, read the criteria, and specifically look for the case you think is most likely and check that the text covers it the way you assume. A view that turns out to be right about the world and wrong about the wording pays nothing.

On a regulated exchange, resolution is an administrative act performed under a published rulebook. The contract specification names a settlement source — an agency’s published figure, an official result, a named body’s certification — and the exchange determines the outcome against it. Where the source is late, revised or silent, the rulebook says what happens, and if you disagree with the determination the route is the exchange’s own process and, beyond it, its regulator.

On an on-chain market, there is no administrator to ask, so resolution runs through an oracle. The common pattern is optimistic: someone proposes the answer and posts a bond, a challenge window opens, and if nobody disputes the proposal within it, the proposal becomes the answer. Polymarket resolves this way, using UMA’s optimistic oracle. If the proposal is disputed, the question escalates to a vote, and the losing side’s bond compensates the winning one.

The two designs fail differently, which is more useful to know than which is better. An exchange’s risk is administrative — a determination you think is wrong, appealed through a process you do not control. An oracle’s risk is one of governance and incentives — the answer is whatever the dispute mechanism concludes, which is designed to track the truth but is not the same thing as the truth.

The resolution date is when the criteria can first be evaluated. The settlement date is when the venue actually credits winning contracts, which may be immediately or may be after a challenge window closes. And the date money is available to withdraw is a third thing again, governed by the venue’s payment rails rather than by the contract.

The gap between the first and the second is where a lot of confusion lives. A market can be effectively decided — trading at 99¢ — and still not have paid anybody, because the mechanism that makes it official has not run yet. During that gap your capital is committed to a position whose outcome you already know, which is a real cost even though nothing looks like it is happening.

It is also where the last remaining risk sits. A contract at 99¢ with a week of challenge window left is not the same asset as a contract at 99¢ that has already settled, and the missing cent is not noise.

Good criteria anticipate the awkward outcomes explicitly. What happens if the event is cancelled, postponed past the deadline, or partially satisfied? What if the named source revises its figure after publication — does the market follow the revision or the first print? What if the source stops publishing altogether? What if two sources disagree?

When the text does not answer these questions, the venue’s general rulebook does, and general rules are blunt: markets get voided, positions get returned at cost, or an extended window gets opened. None of those outcomes is a disaster, but all of them mean your capital was committed for a period and returned without the result you were trading for.

A useful habit is to read the criteria looking specifically for the sentence that resolves the ambiguity you can already see. If it is not there, that is information about the market — and it is often the reason a contract that looks like a near-certainty is trading eight points below where you expect.