Market Resolution

Every prediction market ends the same way. Trading stops, a real-world answer arrives, and each contract that had been drifting somewhere between a penny and a dollar resolves to exactly $1 or exactly $0. That final step is resolution, the process of deciding which outcome actually happened and settling the market accordingly. For most of a contract's life the price is a running probability, readable and reversible. Resolution is where that probability becomes a realized result, and where the hardest questions in these markets tend to surface.

Resolution is also the part newcomers tend to understand least and dispute most. The mechanics that decide a settlement are rarely visible on the trading screen, so a market that seems to have an obvious answer can resolve in a way that surprises the people holding it. This guide follows the resolution process from end to end. It traces a market from close to settlement, identifies who decides the outcome and against what source, shows how the rules are written, compares the two dominant resolution models, walks through a dispute from the inside, and explains why a market sometimes resolves against what everyone watching believed. The event contract itself, and the fixed $1-or-$0 arithmetic that resolution triggers, have their own guide. This one is about how the answer gets decided.

How does prediction market resolution work?

Resolution is a sequence rather than a single moment, and the steps are worth separating because the gaps between them are a common source of avoidable confusion. A market that looks finished on screen may still be days away from settling, and that delay is by design rather than a malfunction.

Three stages structure almost every market:

Close time is the moment the market stops trading, which often precedes the event itself. Determination is when the answer becomes official against whatever source the rules name, and it can trail the event by hours or days when that source is slow, such as an official certification or a final report. Settlement is the last step, when each contract resolves to $1 or $0 and the proceeds reach trader accounts. Kalshi's help center keeps these moments distinct, noting that a market can stay open after an event appears decided and that settlement follows determination rather than the event itself (Kalshi: Market FAQs).

A worked example makes the sequence concrete. Suppose a market asks whether a monthly economic figure will exceed a threshold, and the figure is released on a scheduled morning. The market might close at the release time, so no one can trade on the number once it is known. Determination follows when the agency publishes the official value, which can be minutes later or longer if a revision window applies. Settlement then pays each contract its $1 or $0. The event, the close, the determination, and the settlement are four separate moments, and a trader watching only the news can easily mistake one of them for another.

The lag exists because resolution is only as fast as its source. A market on a weather reading waits for the meteorological agency to publish; a market on an election waits for the result to become official, not merely projected; a market on a corporate event waits for the filing. The regulator's own explainer describes event contracts as settling against a defined outcome from a specified source (CFTC: Understanding Prediction Markets and Event Contracts), and that specified source, not the market's mood or the media's early call, is what determination waits on. The settlement arithmetic that follows is mechanical once the answer is fixed. The interesting part is how the answer gets fixed, which is the subject of the rest of this guide.

Who decides how a prediction market resolves?

Every market names something that will decide it. Some entity or dataset serves as the source of truth, and the resolver's job is to read that source against the question rather than to exercise free judgment. The sources fall into a few recognizable categories:

• Official data. The cleanest markets point at an authoritative public record: a government statistic, an election authority's certified result, a central bank's rate decision, an official scorekeeper's final tally. When such a source exists and is unambiguous, resolution is close to mechanical.
• Credible reporting. Many real questions have no single official datum, so the rules defer to the consensus of reputable sources. A market on whether a public figure takes some action may resolve on agreement among major news organizations, which introduces judgment about what counts as confirmation.
• The rulebook's own determination. Whatever the source, some party applies it to the question and declares the result. That party is the exchange itself on regulated venues, or a decentralized oracle on crypto-native ones, and the difference between those two arrangements is large enough to have its own section below.

The distinction between the source and the resolver matters. The source supplies the facts, and the resolver decides whether those facts satisfy the written criteria. On a CFTC-regulated exchange the resolver is the venue, which settles the market against the source it named when the contract was listed (Kalshi: What are prediction markets?). On Polymarket the resolver is UMA's optimistic oracle, a decentralized system in which participants propose and confirm the outcome rather than a company declaring it (Polymarket: Resolution). Either way, a market resolves to what its criteria and its source support, which is not always what a casual observer would call the obvious answer.

How are a prediction market's resolution rules written?

A market's rules are its rulebook, and reading them is the single most useful habit a trader can build. A well-constructed set of criteria answers, in advance, every question that could plausibly arise about how the market settles. It typically contains a few standard parts:

• The exact question. A single, precise statement of what is being decided, with the key terms defined rather than left to intuition.
• The source of truth. The named dataset, agency, or reporting standard that determination will read.
• The settlement conditions. What outcome resolves the market Yes, what resolves it No, and how the boundary between them is drawn.
• Timing. When the market closes, when the outcome is determined, and what happens if the event is delayed, postponed, or canceled.
• Edge cases. The fine print for awkward situations, such as ties, partial occurrences, a source that goes silent, or an event that happens in an unexpected way.

The forecasting community has thought carefully about what separates a clean rulebook from a leaky one, and its standards transfer directly. Metaculus, which has run public forecasting questions for years, tells its question writers to aim for criteria that are unambiguously resolvable, meaning a reader should be able to agree, both before and after the event, on whether the terms were met (Metaculus: Question Writing Guidelines). Its guidance stresses tight criteria that leave little room for discretion, careful definition of every load-bearing term, and a title that matches the resolution conditions so the headline does not promise something the fine print withholds.

This care matters because a term can feel precise while hiding real ambiguity. A market asking whether one country will "invade" another seems clear until a brief incursion or a proxy action forces the question of what "invade" was defined to mean. Strong criteria settle that in advance by pinning the word to concrete, checkable conditions, and weak criteria leave it to a contested judgment call after the fact. The same discipline explains why a good rulebook can read as almost pedantically detailed. Every clause is there to remove a future argument, and the detail is a property of a market that intends to resolve cleanly.

What is the difference between a resolution committee and an optimistic oracle?

Once a source of truth is named, someone has to apply it, and someone has to handle the case where a trader thinks they got it wrong. Two models dominate, and they map closely onto the two main venue families covered in Venue Types.

The first is the committee model, used by CFTC-regulated exchanges. The exchange operates as the central resolver. Its own team determines the outcome against the stated source and settles the market, and a trader who disagrees raises it through the exchange's support and review process. Authority sits in one accountable place, the operator answers to a regulator, and resolution tends to be fast because no external process has to run. The trade-off is that participants are trusting a single company's judgment and its willingness to correct itself.

The second is the optimistic oracle model, used by Polymarket through UMA. The word "optimistic" names the core idea. Rather than have an authority rule on every market, the system lets anyone propose the outcome, assumes that proposal is correct unless someone challenges it, and escalates to a fuller process only when a challenge appears (UMA: How does UMA's Oracle work?). A proposer posts the answer backed by a bond, a challenge window opens, and if the window passes without a dispute the market resolves as proposed (Polymarket: Resolution). No central committee touches the common case, and the cost of resolving an uncontested market is close to nothing.

Neither model is strictly better, and they distribute the same trust differently. The committee model concentrates it in a regulated operator, and the optimistic oracle spreads it across a set of economically motivated strangers. Both run smoothly on clear outcomes, and both are tested by exactly the same thing, a market whose answer is genuinely contestable.

What happens when a prediction market resolution is disputed?

A dispute is what happens when someone with money at stake believes the proposed outcome is wrong and is willing to pay to contest it. The two models handle that moment differently.

On the optimistic oracle, disputing is a defined, bonded procedure. A challenger posts a bond of their own within the challenge window, which pauses the optimistic resolution and forces the question into a fuller process. Polymarket's implementation runs the escalation in stages, where a first dispute can reset the proposal for another attempt and a second dispute pushes the question all the way to UMA's Data Verification Mechanism (Polymarket: How are markets disputed?; Polymarket uma-ctf-adapter). There the outcome is settled by a vote of the oracle's token holders under a commit-and-reveal scheme, and the economic logic runs through the bonds. The side the vote judges wrong forfeits its bond to the side that was right, so a frivolous dispute is expensive and a correct one is rewarded (UMA: How does UMA's Oracle work?).

On the committee model, a contested resolution routes back through the exchange rather than a public vote. The trader flags the settlement, the exchange reviews the outcome against the source it named, and it either confirms or corrects. The process is private and usually faster, and it rests on the operator's own diligence and its accountability to a regulator rather than on a crowd of stakeholders.

Both paths are exceptional rather than routine. The large majority of markets resolve on an uncontested reading of a clear source, and disputes tend to concentrate in the small share where the criteria and the world fail to line up cleanly. The resolution model a venue uses therefore matters most at the margin, on exactly the contestable markets a careful trader learns to spot in advance.

One principle spans both models and is worth internalizing before trading. Raise doubts while the market is still open. Metaculus advises surfacing an ambiguity while a question can still be forecast, when a clarification can actually change the outcome, rather than after the fact (Metaculus: Question Writing Guidelines). A concern spotted and voiced early can move a listing or a clarification, while the same concern raised after settlement is an appeal against a decision already made.

Why do prediction markets resolve the "wrong" way?

Most markets resolve without drama, and the ones that draw complaints tend to trace back to a small set of causes rather than to bad faith on the venue's part. A market resolves to its written criteria applied to its named source, so what feels "wrong" is usually the distance between that mechanical result and what traders assumed the market was really asking. The recurring patterns are worth recognizing in advance:

• The word problem. A term that felt precise turns out to blur against a messy reality. Whether an event counts as an "invasion," a "recession," or a "ban" can hinge on a definition the casual reader never checked. Researchers who study forecasting call this a specification that is unambiguous yet misaligned, clear on its own terms while failing to capture what the asker actually cared about (Rethink Priorities: Types of specification problems in forecasting).
• Letter over spirit. When the rulebook and common sense diverge, the rulebook wins. A market resolves on the source it named, even if a different source, or the tenor of the day's coverage, points the other way. This can feel unjust to a trader who followed the story rather than the criteria, and it is exactly what the criteria exist to do.
• The timing gap. An outcome that clearly occurred may still not satisfy the market in time. If determination waits on an official certification that lands after the resolution date, or the event happens a day past the deadline, the market can resolve against the outcome everyone watched happen.
• A source that fails or is gamed. Resolution is only as trustworthy as the source it points at. A designated sensor, feed, or report can err, and in some cases it can be deliberately influenced so the market reads a false result, a risk examined in Market Manipulation.
• Unforeseen edge cases. Markets occasionally meet situations their rules never named, such as a tie, a cancellation, a player who does not appear, or a partial occurrence. When that happens, resolution falls back on a judgment call, and judgment calls are where disputes cluster.

The common thread is that a market is a contract, not a conversation. It settles on the terms written down, and the gap between those terms and the shared intuition of the people trading is where almost every "wrong" resolution actually lives. Metaculus's advice to define terms with almost excessive care exists because this gap is so easy to open and so hard to close after the fact (Metaculus: Question Writing Guidelines).

How do you check resolution criteria before trading a prediction market?

Reading the criteria before opening a position is the practical payoff of everything above. The risk being managed has a name, resolution risk, and it refers to the chance that a market settles against you for reasons rooted in how the rules read rather than in whether you were right about the world. A position can have the event correct and still lose on a definition, a source, or a date. Checking the rulebook first is how that risk gets priced in rather than discovered at settlement.

A short pass through the criteria answers most of what matters:

• Read the exact question and the fine print, not the headline. The title is a summary, and the resolution conditions are the contract.
• Identify the source of truth, and ask whether it is a clean official record or a judgment about reporting. The second kind carries more resolution risk.
• Check the determination date and the delay rules. Know what happens if the event is postponed, canceled, or decided after the deadline.
• Know who resolves a dispute, a committee or an oracle, and what contesting a bad resolution would actually require of you.
• Look for the edge case your thesis depends on. If your position rests on a particular reading of a term, confirm the rulebook reads it the same way.

There is a market-reading benefit here too. A price that looks obviously mispriced sometimes reflects a resolution nuance that other traders have already read and you have not, so a surprising number is a cue to reread the criteria before assuming the market is wrong. The broader skill of turning a market page into a judgment is covered in Reading a Market, and the way resolution quirks feed into a market's overall track record is part of Market Accuracy. The habit itself is simple. The rulebook is where a market tells you exactly how it will end, and it is worth reading while you can still act on it.

Related guides