How Prediction Market Contracts Resolve: and What Happens When It Gets Messy

You placed a bet. The event happened. You were right.

Now what?

Most people assume the platform just... pays you. And usually it does. But "resolution" is its own process: the moment a prediction market officially closes and hands out the money. Winning shares pay $1. Losing shares pay $0. Trading stops. Its own rules, its own timeline, its own ways of going sideways.

This is the part of prediction markets that almost nobody explains before you start trading. Here's how it actually works.

The quick answer

When a prediction market resolves, the platform officially declares one outcome the winner and settles all positions. In binary markets (the most common type, where you're betting YES or NO on a single question) winning shares pay $1 per share. Losing shares pay $0. Trading stops. Your balance updates.

That's the core mechanic. Every platform runs some version of it. Where things diverge is in how the platform decides the outcome, when that decision happens, and what counts as proof.

Three things that look the same but aren't

Here's the first thing that trips up beginners: there are three separate moments in resolution, and they don't always happen at the same time.

Close time is when trading stops. After this point, you can't buy or sell shares. The price is frozen.

Determination time (also called outcome time) is when the platform officially decides the result, using whatever evidence source the contract named as its source of truth.

Settlement time is when your balance actually updates. On platforms that hold your money in an account (like Kalshi), this means your cash reflects the result. On blockchain-based platforms (like Polymarket), this is when you can actually redeem your winning tokens for cash.

These three moments are supposed to happen back to back. But Kalshi's own help documentation explicitly warns that close time may not equal determination time, and that settlement only happens after official confirmation from the named source. So if you're staring at a market that closed hours ago and wondering why your winnings aren't there: you might be waiting on a government agency to publish data. The market closed. The judge hasn't ruled yet.

What type of contract you're trading matters

Resolution works differently depending on what kind of contract you're in. There are three main types.

Binary contracts are the YES/NO questions that make up most of what you see on Polymarket and Kalshi. One outcome wins; one loses. Win = $1 per share. Lose = $0. Clean.

Categorical contracts work like binary contracts, but with more than two choices. Think: "Which country will win the most Olympic gold medals?" There are five options. Only one resolves to $1. All the others resolve to $0. Still winner-takes-all, just with more losers.

Scalar contracts are the exception: instead of winner-take-all, your payout slides along a range depending on where the final value lands. If you're trading "What will the Fed funds rate be in December?" your payout isn't $0 or $1. It's somewhere in between, calculated from a formula.

One thing worth knowing: many markets that look like scalar exposure are actually ladders of binary contracts. Instead of one range contract, you get a set of YES/NO questions: "Will it be above 4%? Above 4.5%? Above 5%?" Each question resolves from a single named statistic rather than a continuous calculation, which keeps settlement auditable.

There's also a fourth outcome type you should know about: null resolution. Manifold Markets and Metaculus both support declaring a market void entirely, neither YES nor NO, when the question turns out to be ambiguous, the event never happened, or the contract was poorly specified. On Manifold, this is called "N/A" resolution, and it has an unusual feature: it rolls back all trades, including profits you thought you'd already won.

Who actually decides

Two fundamentally different models govern who makes the call: centralized resolution, where a platform team decides, and decentralized oracle resolution, where a smart-contract system does. Which one you're on affects your dispute options, your timeline, and how much certainty you actually have.

Most explanations skip this. The mechanics of payout are easy. The governance question is where resolution gets interesting: who is the judge, and how can you challenge them.

Centralized resolution means the platform's team makes the call. They look at the contract's written rules, check the named evidence source, and determine the outcome. Kalshi operates this way. Its markets team reviews outcomes when resolution criteria are met, and its formal rulebook describes an Outcome Review Committee that can be convened for contested or complex outcomes. That committee's determinations are final, and they're made within 24 hours of review being initiated.

This model is familiar. It's how sportsbooks work. Someone official reads the scoreboard and pays the winners. The difference is that prediction markets trade on all kinds of events, not just sports results. So the "scoreboard" is sometimes a government data release, a court ruling, or a news report. When that source is delayed or ambiguous, so is your payout.

Decentralized oracle resolution means the platform uses a smart-contract system, a set of code rules that run automatically on a blockchain, rather than a human team making judgment calls. Polymarket uses this model. When a market is ready to resolve, anyone can submit a proposed outcome. That proposal is accepted automatically after a 2-hour challenge window. Unless someone disputes it. If disputed, the question escalates to a vote by holders of UMA tokens (a governance token for the oracle system). Votes take roughly 48 hours. The entire disputed timeline runs 4–6 days.

One number worth knowing: UMA's oracle has settled over 22,000 outcome assertions. Only 1.67% were disputed. For most markets, the process is fast and quiet. The drama lives in the 1.67%, and that 1.67% tends to be the largest, most contentious markets on the platform.

The underlying concept connecting both models is something traders call verification risk. You're not trading on what actually happened in the world. You're trading on what the contract's named evidence source will confirm, in the way the contract requires. That gap — between reality and provability — is where disputes are born.

The part where resolution gets genuinely funny

Prediction markets will put a contract on anything. The resolution process, however, is completely humorless. It applies the same contract logic to every question, no matter how strange. Sometimes the collision between contract language and reality produces results that are harder to predict than the underlying event.

The case of the suit that wasn't (maybe): In mid-2025, Polymarket ran a market asking whether Volodymyr Zelenskyy would wear a suit before a specific date. This sounds like a joke. It generated over $160M in trading volume and ended in a formal dispute. The question wasn't whether Zelenskyy wore something formal. Under the contract, the question was whether he was photographed or videotaped "wearing a suit" within a specific window, as confirmed by "a consensus of credible reporting." The dispute turned on whether a particular jacket met the definition of a suit under the contract's terms.

The tweet count problem: Markets on Elon Musk's tweet volume have repeatedly run into the same issue: the contract has to define what a "tweet" is. Do deleted tweets count? Do retweets? Quote-tweets? What if the API that counted them changes? What timezone? These aren't exotic edge cases invented by critics. They're the predictable result of trying to make a human behavior — posting online — auditable by a contract that must produce a single verifiable number.

The resolution criteria for the Second Coming: "Will Jesus Christ return before 2027?" is an actual Polymarket contract. The market exists. What makes it interesting isn't the theology. It's that someone had to write resolution criteria specifying what would constitute verifiable confirmation of this event.

The lesson isn't that these markets are frivolous. It's that resolution criteria are a legal contract, and legal contracts applied to the real world always find the edge cases the drafters didn't anticipate.

The six ways resolution goes wrong

Resolution problems cluster into six categories. Knowing these before you trade is the difference between losing a bet and losing a bet you thought you'd won.

1. Ambiguity risk — The contract language doesn't cleanly map to the real-world outcome. The Zelenskyy suit market is the canonical example. The question seemed obvious. The contract's definition differed from what most traders assumed.

2. Buried exception risk — The platform has a rule that isn't surfaced in the user-facing contract summary. The outcome you predicted occurs. The exception fires anyway.

3. Source risk — The named evidence source is delayed, revised, or in extreme cases, influenced. In March 2026, a journalist named Emanuel Fabian reported that an Iranian missile had landed near Beit Shemesh on March 10. His reporting became the primary evidence source for a Polymarket resolution. He received threats and bribe attempts from traders trying to influence how his article would be interpreted for market resolution.

4. Timing risk — Close time and determination time are not the same. Markets that rely on government data releases, election certifications, or official statistics can sit in limbo for hours or days after trading has stopped.

5. Governance risk — In oracle or committee-based systems, high-stakes and highly interpretive disputes can become legitimacy crises. In March 2025, a ~$7M bet on a Ukraine mineral deal triggered a public conflict between Polymarket's community and UMA's oracle stakeholders, with allegations of governance manipulation on both sides.

6. Null outcome risk — The market resolves void. On Manifold, this means trades are rolled back entirely. Profits disappear. This is the fairness mechanism for badly written markets. It's also how you lose money you thought you'd won.

The Khamenei case: what buried exception risk looks like in court

In January 2026, Kalshi opened a market titled "Ali Khamenei out as Supreme Leader?" The rule displayed on the market page: if Khamenei leaves office before March 1, 2026, the market resolves YES.

The market accumulated $54 million in trading volume over the following weeks. US-Israeli strikes under Operation Epic Fury killed Khamenei on February 28, 2026, the day before the first expiration date.

Kalshi invoked a "death carveout provision." The clause specified that if a leader leaves office solely because they died, the market resolves at the last traded price, not at $1. Traders holding YES positions received fractional payouts, not the full dollar per share they expected.

The carveout was not in the user-facing rules summary. Kalshi's CEO, Tarek Mansour, acknowledged on social media that disclosures had been "grammatically ambiguous" and announced that Kalshi would reimburse all fees and net losses out of pocket.

A class action lawsuit was filed on March 5, 2026 (Risch v. KalshiEX LLC, C.D. Cal. No. 2:26-cv-02390), covering all US holders of YES positions across every expiration date when trading halted. The lawsuit describes a "predatory scheme," alleges Kalshi kept the market open and actively promoted it as military strikes were accumulating, and argues that for an 85-year-old autocratic leader with a US naval fleet on Iran's doorstep, death was "the most likely, and in many cases the only realistic, mechanism" for him to leave office.

The user-facing rule said "leaves office." The technical rule said "but not if he dies." Those two sentences are not the same contract.

The case is ongoing.

What to check before you trade

Most traders who get burned by resolution didn't lose because they predicted wrong. They lost because they didn't read the right document.

The title is marketing. The rules are the contract. On Polymarket especially, these are different texts. Before you buy a position, find the rules section and read it. Not the headline. The rules.

Find what counts as proof. The contract should name a specific agency, publication, or data feed. If the source is vague ("credible reporting" or "consensus of media"), you are exposed to the same kind of interpretation fight that swallowed the Zelenskyy suit market.

Look for carveouts. Words like "except," "unless," "provided that," and "solely due to" are where the buried exceptions live. If you can't find them in the summary, look in the full rules documentation. Kalshi admitted its own death carveout was "grammatically ambiguous" after $54M had traded on it.

Know who resolves and what they can override. A tokenholder community on Polymarket, a formal committee on Kalshi, the market creator on Manifold. These are not equivalent. What happens in a close call differs on each platform.

None of this guarantees you've found every exception. But it closes most of the gap between thinking you understand a contract and actually understanding it.

Key takeaways: prediction market resolution

Resolution is not a technicality. It's the entire point: the moment when "I think this will happen" turns into "I was right and here's the money." Almost everything that goes wrong in prediction markets goes wrong here. The rule you didn't read, the source that didn't confirm, the carveout that didn't appear in the summary.

The good news: this is learnable. Polymarket publishes its full dispute mechanics. Kalshi files its rulebook with the CFTC. The information exists. Read the rules. Find the source. Check for exceptions. Know who resolves and what they can override.

The market doesn't care what you assumed. The contract does.