How do prediction markets resolve? Settlement sources, disputes, and what happens when the rules are unclear
The rules page is the contract
Every market on a regulated prediction market is a short legal document dressed as a question. Kalshi's help centre puts it directly: "Each market operates under its own set of regulations which are documented on the markets page." The Rules Summary spells out what must happen for the contract to resolve Yes and names the outcome verification source, which Kalshi calls "an authoritative point of reference." Polymarket's documentation says the same thing in different words: each market defines its resolution source, the end date on which it becomes eligible for resolution, and how edge cases are handled, and "the rules, not the title, determine the outcome."
That sentence is the single most important thing on this page. A market titled "Will Cardi B perform at the Super Bowl?" is not resolved by whether most people would say she performed; it is resolved by whatever the rules say "perform" means, read against whatever the named source reports. When traders lose money on a resolution they consider unfair, the cause is almost always a gap between what the title implied and what the rules text said, and the rules text was available before the first trade. The practical habit this creates is simple: before buying, read the rules, find the source, and ask what the source will actually publish and when.
How Kalshi settles a market
Kalshi is a designated contract market regulated by the Commodity Futures Trading Commission, and its resolution process is centralised. Three documents describe it. The help centre says Kalshi's markets team reviews all markets and that outcomes are determined "when decided that resolution criteria has been met"; it also warns that a market remaining open does not by itself mean the criteria have or have not been met. The developer documentation says that once an outcome is determined, positions are "automatically resolved and funds transferred," that winning Yes or No contracts pay $1 each, that only net positions settle, and that markets "typically settle shortly after expiration," with timing that varies by market type, data source availability and whether manual review is needed. And the contract terms Kalshi files with the CFTC set the legal frame: the expiration value is "the value of the Underlying as documented by the Source Agency," the settlement date is "no later than the day after the Expiration Date, unless the Market Outcome is under review," revisions to the underlying published after expiration are not counted, and if an expiration value cannot be determined on the expiration date, Kalshi has the right to determine payouts under its rulebook.
Two features of that design are worth stating plainly. First, the source is fixed per market and is usually an official body: a league for a game result, the Bureau of Labor Statistics for a jobs number, the Federal Reserve for a rate decision, the National Weather Service for a temperature. Where a market has a clean official source, settlement is mechanical and fast. Second, there is no open dispute process. A trader who disagrees with a settlement can write to support, flag the market, or complain to the CFTC, but the exchange decides, and the contract terms give it discretion to open an outcome review before settlement and to determine payouts when the source fails. That is the trade-off of a regulated central venue: speed and a named authority in exchange for a decision you cannot appeal inside the platform.
| Step | Kalshi | Polymarket (UMA-resolved markets) |
|---|---|---|
| Who writes the rules | Kalshi; contract terms filed with the CFTC | Polymarket; rules shown under the order book |
| Who determines the outcome | Kalshi's markets team, reading the named source | Any user who proposes an outcome and posts a bond |
| Bond to propose | None; traders may request settlement but cannot propose | About $750, forfeited if the proposal is wrong or too early |
| Window to object | None open to the public; Kalshi may open an outcome review | 2 hours; anyone may dispute by posting a matching bond |
| Final decision if contested | Kalshi | UMA token-holder vote after a second disputed proposal |
| Time to pay, uncontested | Typically shortly after expiration; no later than the day after under the contract terms | About 2 hours after the proposal |
| Time to pay, contested | Until the review ends | 4 to 6 days: 24 to 48 hours of debate, then about 48 hours of voting |
| If the outcome cannot be determined | Kalshi determines payouts; has settled at the last traded price | Voters may return "unknown" and the market pays 50 cents a token |
| Payout to the winning side | $1 per contract | $1 per token |
How Polymarket resolves a market
Polymarket's international platform does not employ a settlement desk. It uses UMA's optimistic oracle, a smart-contract system in which the default assumption is that a proposed answer is correct unless someone pays to say otherwise. The process has four stages, and the numbers below are from Polymarket's resolution documentation as of October 2026.
First, a proposal. Once a market's end date has passed, anyone can propose the outcome by posting a bond, typically $750, and submitting it to the oracle. A correct, undisputed proposal returns the bond plus a reward; a wrong or premature one forfeits the whole bond, which is why the documentation warns users to propose only if they are confident. Second, a challenge period of two hours. If nobody disputes, the proposal is accepted and the market resolves roughly two hours after it was made. Third, a dispute. Anyone who disagrees posts a counter-bond equal to the proposer's, and the first dispute simply starts a new proposal round. If the second proposal is also disputed, the question goes to a vote. Fourth, the UMA vote: a debate period of 24 to 48 hours in which evidence is posted in UMA's public channels, then about 48 hours of voting by holders of UMA's token. The vote can return the proposer's answer, the disputer's answer, "too early," or, rarely, "unknown," in which case the market resolves 50/50 and every token on both sides redeems for 50 cents. In each case the losing bond is split, with half going to the winner.
Polymarket also reserves the ability to publish an "additional context" note on a market, recorded on-chain, that voters are expected to consider. The documentation says such a note cannot change the fundamental intent of the question. In practice, as the examples below show, clarifications are the most contested part of the system, because they are issued after trading has begun and sometimes after the event itself. Separately, Polymarket's short-duration crypto Up/Down markets do not go through UMA at all: they compare a Chainlink time-weighted price at the start of the window with one at the end, and resolve Up if the final price is equal to or above the starting price, with no proposal or dispute step.
One more distinction matters for US readers. Polymarket's CFTC-registered US exchange, which operates under its own rulebook, does not route settlement through UMA; reporting on its rules describes central settlement by a markets team, closer to Kalshi's model than to the international platform's. The two Polymarkets share a brand and a market list but not a resolution mechanism, and we set out the regulatory difference in is Polymarket legal in the US.
A worked example: one market, two venues, two answers
The cleanest illustration of how resolution rules bite is the halftime show at Super Bowl LX in February 2026. Both venues listed a market on whether Cardi B would perform. She appeared on stage during Bad Bunny's set, danced, and may or may not have sung; reporting at the time said she could have been mouthing the words. Here is what each venue did with the same facts, and what a trader holding 100 Yes contracts bought at 40 cents would have received.
| Kalshi | Polymarket (international) | |
|---|---|---|
| Rule text | Singing or dancing counts as a performance; background dancing does not | Yes if she performed "live and in person"; a later additional-context note said participation without singing qualified |
| Volume on the market | More than $47 million across the "who will perform" series | About $10 million on the halftime series, more than $5 million on Cardi B |
| What the venue decided | Outcome could not be clearly determined; market settled at the last traded price, which Kalshi called fair value | Resolved Yes after a UMA process in which traders disputed the timing of the clarification |
| Payout per contract | Yes holders 26 cents, No holders 74 cents | Yes holders $1, No holders $0 |
| 100 Yes contracts bought at 40 cents ($40 plus fee) | Receive $26, a loss of about $14 plus the fee | Receive $100, a profit of about $60 minus fees |
| Recourse | At least one user filed a complaint with the CFTC | Dispute by bond, then token vote |
Neither venue kept the money; Kalshi said it retained none of the funds and Polymarket's payout went to Yes holders. The point of the example is not that one venue was right. It is that identical facts produced a 74-cent swing per contract because the two rulebooks handled ambiguity differently: one by declining to pick a side and settling at the market's last estimate, the other by letting a clarification and a vote pick a side. A trader who had read both sets of rules would have known before kick-off that the Kalshi contract carried a provision for an undeterminable outcome and that the Polymarket contract could be clarified after the fact.
Where resolution goes wrong, and the provisions that catch it
Most markets settle without incident. The disputes that reach the press cluster into four patterns, each of which maps to a specific rule you can check in advance.
The source published late, or outside the window. A recurring Polymarket market in 2026 asked whether Strategy, the bitcoin-holding company, would sell any bitcoin by the end of May. It sold 32 bitcoin during the last week of May and disclosed the sale in a regulatory filing on 1 June. Yes briefly traded near 80 cents on the filing; Polymarket then posted a note that confirmation outside the market's timeframe did not qualify, and after two disputed proposals and a token vote the market resolved No on 3 June. The lesson is that an event-date rule and a confirmation-date rule are different rules, and the clarification mechanism can decide which one applies. Kalshi's contract terms address the same risk from the other direction: the expiration value is what the source documents at expiration, and later revisions do not count.
The source was named incorrectly. In 2025 a Kalshi football market listed a school district as its settlement source, which Kalshi described as a display error and corrected. Rules pages are written by people and occasionally contain mistakes; the correction process on Kalshi is internal, and on Polymarket it is the additional-context note.
The rules did not anticipate the event. A soccer market on Kalshi with no draw option went to extra time; a season win-total market in January 2026 was initially settled in a way that returned stakes rather than paying winners, then corrected after press coverage. Kalshi's provision for an outcome that cannot be determined is what produced the 26/74 split above, and traders should assume any market whose question has a third possible answer may end up there.
The decision-maker had a stake. This is the structural criticism of the UMA model. A Wall Street Journal analysis cited in Galaxy Research's review of the Strategy dispute found that in most disputed Polymarket markets more than half the votes came from the ten largest wallets, and that about one dispute in five involved a voter with a position in the contract. UMA's design assumes that losing a bond disciplines bad proposals and that token holders vote honestly to protect the token's value; the concentration figures are the reason some traders treat large disputed markets as carrying resolution risk on top of event risk. Kalshi's structure avoids that problem by concentrating the decision in the exchange, which creates the opposite objection, voiced in at least one state attorney general's complaint, that the process sits entirely inside the company whose fees depend on the market.
Five questions to ask before you buy
Resolution risk is cheap to check and expensive to ignore. Before taking a position on either venue, find answers to these in the rules text. What exactly is the question, with every defined term read literally rather than as a headline? Which source settles it, does that source publish on a schedule, and what happens if it is late, revised or silent? What is the expiration or end time, and is it the time of the event or the time of confirmation? Is there a provision for an undeterminable outcome, and if so does it pay $1, $0, the last price or 50 cents? And can the rules be clarified after listing, and if so by whom? Markets on official statistics, scheduled votes and league-certified results score well on all five. Markets on what a person will say, whether something counts as a performance, or whether an announcement will be made by a date score badly, and their prices should carry a discount for it.
What this guide does not cover
This is a guide to the published resolution mechanics of the two largest venues, not to every prediction market or to every market type. It does not cover Kalshi's perpetual futures, scalar or range markets that settle to a value between $0 and $1, or the settlement floor Kalshi applies to some markets; the details of UMA's governance and token economics; the internal rulebook of Polymarket's US exchange beyond what has been publicly reported; or how to profit from resolution disputes, which we do not recommend attempting. The Kalshi figures are from its help centre, developer documentation and CFTC-filed contract terms, and the Polymarket figures from its resolution documentation, all read in October 2026; both venues change their rules, and the live pages govern. Nothing here is legal, tax, financial or betting advice.
Frequently asked questions
Who decides how a prediction market resolves?
On Kalshi, the exchange's own markets team determines the outcome by checking the settlement source named in each market's rules, and the exchange can open an outcome review before settlement. On Polymarket's international platform, anyone can propose an outcome to the UMA optimistic oracle by posting a bond; if nobody disputes it within two hours it stands, and if it is disputed twice, holders of UMA's token vote. Polymarket's CFTC-registered US exchange settles centrally under its own rulebook rather than through UMA.
How long does it take for a prediction market to pay out?
Kalshi says markets typically settle shortly after expiration, and its contract terms set the settlement date no later than the day after the expiration date unless the outcome is under review. An undisputed Polymarket market resolves about two hours after a proposal is made; a disputed one takes four to six days because of the debate period and the 48-hour token vote. Winning contracts pay $1 each on both venues.
What happens if the settlement source is wrong or unavailable?
Kalshi's contract terms say that if an expiration value cannot be determined on the expiration date, Kalshi has the right to determine payouts under its rulebook, and in at least one documented case it settled an unresolvable market at the last traded price rather than at $1 or $0. Polymarket's rules allow a market to resolve 50/50 if UMA voters find the outcome unknowable, with each token redeeming for 50 cents. Revisions to a statistic published after expiration do not change Kalshi's expiration value.
Can you dispute a prediction market resolution?
On Polymarket's UMA-resolved markets, yes: anyone can dispute a proposal during the two-hour challenge period by posting a counter-bond, typically $750, and the dispute is settled by a token-holder vote. On Kalshi there is no open dispute mechanism; traders can contact support or flag a market, and the exchange may open an outcome review at its discretion, but the decision is Kalshi's. Complaints about a CFTC-registered exchange can also be filed with the CFTC.
Related
- How does Kalshi work? Contracts, prices, fees and payouts
- Is Polymarket legal in the US? Two Polymarkets, one federal licence
- What happens to a tennis contract if a player retires mid-match?
- Is Kalshi legal in your state? The 50-state tracker
- Sportsbook vs exchange price comparison tool
Sources
- Kalshi Help Center, How are markets settled?, for the per-market rules, the outcome verification source and the role of the markets team; and When do markets close and settle?, for closing and settlement dates and early settlement.
- Kalshi, Market settlement, developer documentation, for the $1 payout, netting and settlement timing.
- Kalshi contract terms filed with the CFTC, rule certification, July 2026, for the expiration value definition, the settlement date, post-expiration revisions, the outcome review provision and the undeterminable-value provision.
- Polymarket, Resolution, for the proposal bond, the two-hour challenge period, the dispute and voting timeline, the 50/50 outcome, additional context notes and Chainlink-resolved markets; and Polymarket Help, How are prediction markets resolved?.
- Front Office Sports, Cardi B is a cautionary tale for prediction markets, February 2026, for the halftime-show resolution on both venues, volumes and the CFTC complaint.
- Galaxy Research, Strategy bitcoin sale: Polymarket resolution dispute, June 2026, for the May 2026 dispute timeline and the voter concentration figures; Galaxy disclosed a position in that market.
- DeFi Rate, How contracts and dispute resolution work at prediction markets, for the catalogue of earlier Kalshi settlement disputes and the description of Polymarket US settlement.
18+ only. Prediction market contracts carry real financial risk and you can lose everything you pay for a position. This page explains published resolution rules; it is not legal, tax, financial or betting advice. If gambling stops being fun, help is available at 1-800-GAMBLER.