How Kalshi and Polymarket Settle Event Contracts (and Disputes)

A trader buys YES. The event happens. They get paid. That’s how most traders think event contract settlement works — and for the majority of markets, it does. But when the outcome is ambiguous, when contract language doesn’t match what actually happened, or when two prediction markets are looking at the same event and reaching different conclusions, settlement becomes the most important part of the trade.

It’s also the part with the least protection if something goes wrong. This article breaks down how settlement actually works, where it’s broken and what traders need to know.

What settlement actually means

Every event contract follows the same lifecycle: a market opens, you buy and sell YES and NO contracts that sum to roughly $1, the market closes to new trades at expiration, and then a resolution process determines which side won. Winners receive $1 per contract. Losers receive nothing.

Settlement is what happens after resolution — the actual transfer of funds from the losing side to the winning side. On Kalshi, settlement typically occurs within a few hours after resolution is determined. On Polymarket, settlement occurs after the UMA oracle process completes, which takes a minimum of two hours if uncontested but can stretch to days or weeks if a dispute escalates.

The critical point you may not have considered is that settlement is not automatic. Someone or something has to determine the outcome first, and that’s where the two dominant platforms take fundamentally different approaches.

Before a contract settles, it has to be matched. Here’s how prediction market order books pair buyers and sellers.

How Kalshi determines outcomes

Kalshi operates a centralized resolution model. Every market you trade on Kalshi lists named “Source Agencies” in its contract terms, which are filed with the CFTC as part of the exchange’s self-certification process. These are the entities whose data determines whether you get paid.

The source agencies vary by market category:

The actual decision, though, sits with Kalshi’s internal markets team. As the platform’s help center puts it, outcomes are determined “when decided that resolution criteria has been met.” Traders can submit a “Request to Settle” through the platform interface, but this functions as a suggestion, not a binding action. Kalshi’s team makes the final call.

When outcomes are ambiguous or disputed, Kalshi’s rulebook provides for an Outcome Review Committee — a committee of the board that can make binding determinations. The platform can also invoke Rule 6.3(c), which allows it to settle a market at its last traded price if the outcome is deemed unresolvable. That rule became nationally relevant during the Super Bowl.

There is no formal arbitration process, no independent appeal mechanism, and no external ombudsman.

As one trader on Manifold Markets documented after losing money on an incorrectly resolved Oscars market, the only recourse is informal: “Yell at them in discord.”

Kalshi settlement disputes

The centralized model means that when Kalshi’s team makes mistakes, there’s no structural check to catch them unless traders like you raise enough noise on social media.

Kalshi’s internal team determines outcomes, and when errors occur, corrections require external pressure — typically social media exposure — rather than any built-in review mechanism.

Kalshi appears to be addressing the gap. The company expanded its surveillance and enforcement framework — including an independent advisory committee, external analytics partnerships, and a dedicated head of enforcement — and was actively hiring for a dedicated Sports Operations team focused on market determinations, edge case handling, and automated settlement — roles that didn’t exist during most of the incidents above.

The Massachusetts Attorney General’s complaint against Kalshi framed the structural concern directly: Kalshi “writes the rules for the contract, determines the basis for settlement,” and the entire process “functions entirely within Kalshi’s corporate structure and does not serve as an independent intermediary.”

How Polymarket determines outcomes

Polymarket uses a decentralized resolution model built on UMA’s Optimistic Oracle. The system is designed so that no single entity decides outcomes — at least in theory.

The resolution flow follows a structured escalation:

1. Proposal. When a market reaches its resolution date, anyone can propose an outcome by posting a $750 USDC bond.
2. Challenge period. A two-hour window opens. If nobody disputes the proposal, the outcome is accepted as correct and the market settles.
3. First dispute (auto-reset). If someone disputes the proposal, the system resets — a new proposal is requested and the first dispute is effectively ignored. This prevents frivolous disputes from slowing down straightforward resolutions.
4. Second dispute (DVM escalation). If the second proposal is also disputed, the question escalates to UMA’s Data Verification Mechanism (DVM), where UMA token holders vote on the correct outcome.
5. DVM vote. Voting takes 48–96 hours. Votes use a commit/reveal scheme to prevent collusion, and voters stake UMA tokens that can be slashed if they vote against the majority. The system is built on a Schelling Point assumption: independent voters are incentivized to report the truth because they expect others to do the same.

Roughly 98.5% of markets resolve at the Optimistic Oracle layer without ever reaching the DVM. For straightforward outcomes — who won a football game, what was the closing price of Bitcoin — the system works efficiently. The problems surface when outcomes involve interpretation.

Polymarket US and Polymarket International have two different settlement processes. On the US version, the Markets Team is directly responsible for determining outcomes. On the International version, the Markets Team provides clarifications, but the UMA Oracle performs the actual resolution. This means the same market can follow a different resolution process depending on which version of the platform you’re using.

Structural risks in the UMA system

UMA’s voting mechanism is token-weighted, meaning influence scales with holdings rather than following a one-person-one-vote model. DefiLlama founder 0xngmi originally flagged this imbalance in 2025, and the gap has widened since. UMA’s circulating market cap currently sits around $44 million (per CoinGecko, Feb 2026), meaning 51% control costs roughly $22 million. Polymarket’s total value locked, meanwhile, has grown to approximately $330 million (per DefiLlama, Jan 2026). That’s a 15:1 ratio between the capital at risk on the platform and the cost of controlling the oracle that settles it.

Your positions are secured by a system that costs a fraction of the money riding on it to compromise.

The $750 bond requirement creates an additional barrier. If you believe a market has been resolved incorrectly, you need to risk $750 to dispute it — a sum that’s enough to discourage most retail traders from challenging outcomes even when they have evidence the resolution is wrong.

Polymarket clarifications function as binding rules within the UMA system. UMA voters have historically followed Polymarket’s stated position, and according to one analysis, UMA has never overturned a Polymarket clarification. This creates a dynamic where Polymarket’s internal team retains meaningful influence over outcomes despite the decentralized resolution infrastructure.

Polymarket resolution controversies

The UMA system has been tested by several high-profile disputes:

One Trustpilot reviewer summarized the frustration shared across multiple incidents: outcomes may be resolved based on “the spirit of the market” or a strict interpretation of the rules, and which approach is applied is unpredictable.

When the same event settles differently

The most consequential risk if you trade across platforms is settlement divergence, where Kalshi and Polymarket resolve what appears to be the same event in opposite directions.

The Cardi B Super Bowl halftime market is the clearest recent example. Both platforms asked whether Cardi B would perform during the halftime show. Both saw significant volume ($47.3 million on Kalshi, over $10 million on Polymarket, per DeFi Rate’s volume tracker). After Cardi B appeared on stage, danced, and appeared to mouth lyrics during Bad Bunny’s set, the platforms reached opposite conclusions.

Kalshi determined the outcome was ambiguous and invoked Rule 6.3(c), settling at the last traded price. YES holders received $0.26 per contract. NO holders received $0.74. Polymarket resolved YES at $1, paying out YES holders in full.

The divergence wasn’t caused by different information — both platforms had access to the same broadcast footage. It was caused by different resolution criteria, different interpretation standards, and different institutional incentives.

Kalshi’s rules distinguished between dancing that qualifies as performing and dancing that doesn’t. Polymarket’s rules relied partly on a “consensus of credible reporting,” and most major media outlets described Cardi B as having performed.

If you’re running arbitrage — long YES on one platform, short YES on the other — settlement divergence turns what looks like a risk-free spread into a potential total loss.

Platform fees compound the problem: Kalshi and Polymarket use different fee structures that can erode thin arbitrage margins even before divergence risk enters the picture.

You can lose on both sides if the platforms resolve in the same direction, or produce wildly different P&L if they resolve in opposite directions. The assumption that “same event = same outcome” breaks down when resolution criteria, timing windows, and decision-making processes differ across platforms.

Centralized vs decentralized settlement: the real tradeoff

Neither model has earned a clean track record.

The honest framing is not regulated vs unregulated, or centralized vs decentralized. It’s that centralized settlement carries operational risk — humans make mistakes and the platform has structural conflicts of interest in adjudicating outcomes on its own exchange.

If you’re trading on an app that isn’t Kalshi or Polymarket, your settlement process is determined by the underlying exchange, not the app. PrizePicks, Sleeper, Coinbase, and Robinhood all route trades through Kalshi as futures commission merchants, so they inherit Kalshi’s resolution criteria, source agencies, and dispute process (or lack of one). DraftKings Predictions and FanDuel Predicts route through CME Group. Fanatics Markets runs on Crypto.com’s Derivatives North America (CDNA) exchange. For a full breakdown of which apps route through which exchanges, see our list of prediction markets. The app you’re trading on doesn’t determine how your market settles — the exchange it routes through does.

Decentralized settlement carries governance risk — token economics can be gamed, resolution criteria can be reinterpreted after the fact, and the cost of disputing outcomes is prohibitive for most retail traders.

If either platform gets your market wrong, you don’t have great options on either side.

What to check before holding through settlement

Resolution criteria deserve the same attention as your entry price. For any market you plan to hold through expiration:

• Read the actual contract language — not the market title. On Kalshi, resolution criteria are in the contract terms (CFTC-filed PDFs and the market page). On Polymarket, they’re in the market description and the ancillary data stored on-chain.
• Identify the resolution source — Kalshi names specific source agencies for each market. Polymarket names resolution sources in the market description. If the source is ambiguous or undefined, treat that as a risk factor.
• Compare cross-platform criteria — if the same market exists on both platforms, read the resolution language side by side. Differences in wording, even subtle ones, can produce different outcomes when real-world events don’t fit neatly into YES/NO categories.
• Know which resolution process applies — on Polymarket, whether you’re on the US or International version determines whether the Markets Team or the UMA Oracle resolves your market. On Kalshi, resolution is always handled by the internal team with the Outcome Review Committee as a backstop for disputed cases.
• Understand your recourse — on Kalshi, if the team determines the outcome incorrectly, your options are email support and Discord. On Polymarket International, you can dispute a resolution by posting a $750 USDC bond and going through the UMA process. Neither platform offers formal arbitration.
• For arbitrage positions, factor in divergence risk — settlement divergence is not theoretical. It happened with the Cardi B Super Bowl market. Price spreads between platforms may reflect genuine differences in resolution probability, not just market inefficiency. Use a cross-platform arbitrage calculator to model net returns after fees before committing capital.

Settlement is where the trade actually happens. Everything before it is positioning.