Guide

Will this market resolve the way you think?

Q: How does Polymarket decide the outcome of a market?

Polymarket resolves markets through UMA's optimistic oracle. A proposer asserts the outcome with a bond; if no one disputes it within a challenge window, it stands and the Conditional Token Framework contracts on Polygon pay out the winning side. If it is disputed, UMA token holders vote to settle it. The final payout is recorded on-chain, which is the closest thing to ground truth.

Q: Can a Polymarket market resolve unexpectedly?

Yes. Ambiguous wording, edge cases the question did not anticipate, timing details, and disputes can all push a resolution somewhere traders did not expect from the price. The market price reflects probability, not certainty about how the rules will be applied — those are different risks.

Q: Do Polymarket and Kalshi resolve the same event the same way?

Not always. Kalshi is a CFTC-regulated exchange that settles against its own published contract rules; Polymarket settles via the UMA oracle on-chain. Two markets that look like the same question can carry different wording and different resolution sources, so they can settle differently. That gap is settlement-mismatch risk, and it matters for anyone pair-trading across venues.

Q: What is settlement risk in prediction markets?

Settlement risk is the chance that a market resolves differently from what the price and the obvious reading of the question imply — because of ambiguous terms, an oracle dispute, an edge case, or a cross-venue rule difference. It is distinct from being wrong about the event itself: you can call the event correctly and still lose on resolution.

Q: How do I get prediction-market resolution data by API?

Idmon exposes GET /api/resolutions for recent resolutions with a calibrated, on-chain-anchored truth label and dispute flags, and GET /api/resolutions/{conditionId} for one market's full label plus the per-source evidence trail. Both are under https://api.idmon.io with a Bearer key. See the docs at idmon.io/docs.

21 June 2026 · 8 min read

The price of a prediction market tells you the odds. It does not tell you whether the market will settle the way you assume. Resolution — the moment a market is declared YES, NO, or void — is where the money is actually won or lost, and it's where the surprises live. You can call an event perfectly and still lose, if the rules resolve somewhere you didn't expect.

Resolution is not the same as the price

Traders spend almost all their attention on whether the event will happen, and almost none on how the market is worded and who decides the outcome. Those are two separate risks. The first is about the world; the second is about the contract. A market trading at 0.92 is telling you the crowd thinks the event is very likely — it is not telling you that the resolution rules are unambiguous, or that an oracle dispute can't send it the other way.

How Polymarket actually resolves

Polymarket settles markets through UMA's optimistic oracle. Once a market closes, a proposer asserts the outcome and posts a bond. If nobody disputes that assertion within a challenge window, it's accepted and the Conditional Token Framework (CTF) contracts on Polygon pay out the winning side. If it is disputed, the question escalates to a vote of UMA token holders, who settle it.

The useful property here is that the final outcome is recorded on-chain as a payout vector — the contract literally encodes which side won. That on-chain record is the closest thing the ecosystem has to ground truth, and it's verifiable by anyone. The catch is the path to get there: the optimistic model assumes someone will dispute a bad assertion, and disputes hinge on how the question was written.

How Kalshi resolves — and why venues disagree

Kalshi works differently. It's a CFTC-regulated exchange, and it settles each contract against its own published rules and designated sources of truth, determined centrally rather than by an on-chain oracle. Both models can be correct and still disagree with each other, because the thing that resolves isn't the event — it's the contract, and two contracts about the "same" event can carry different wording, different cut-off times, and different sources.

This is settlement-mismatch risk, and it's the trap in naive cross-venue arbitrage. Two markets that look identical — "Will X happen?" on each platform — can settle on opposite sides because one used a stricter definition or an earlier deadline. Anyone trading the spread between venues is implicitly betting that the resolutions agree, which is an assumption worth checking rather than assuming.

Where settlement risk hides

Ambiguous wording. "By the end of the month" — calendar month or 30 days? Announced, or merely reported? Small wording choices decide real money.
Edge cases the question didn't anticipate. A candidate drops out, an event is postponed, a definition turns out to be contested. The rule has to be applied to a situation it wasn't written for.
Disputes. An optimistic oracle is only as good as the people watching it. A contested assertion can flip an outcome the price treated as settled.
Timing. When exactly does the clock stop? Markets have resolved on a technicality of when something counted, not whether it happened.

How to measure resolution truth

Most data tools record only the final winning outcome — a single string, with no indication of how solid it is. That throws away the information that actually matters for settlement risk. A more honest resolution record carries four things:

An on-chain anchor. Read the CTF payout vector directly from Polygon, rather than trusting a scrape that might be stale or wrong.
Multi-source agreement. Compare what the on-chain settlement, the platform's own API, and the final market price each imply. When they agree, confidence is high; when they don't, that's a flag.
A calibrated confidence. Not just "YES", but a calibrated P(label correct) — a number you can actually threshold on, fitted against a history of resolved markets.
Dispute and review flags. An explicit marker for resolutions that were contested or that need a human look, so you can surface settlement-risk cases instead of being surprised by them.

Getting resolution truth by API

Idmon builds exactly this record — on-chain-anchored, multi-source-fused, calibrated — and exposes it as two endpoints:

GET /api/resolutions — recent resolutions with the calibrated truth label, settlement state, and dispute/review flags. Pass reviewOnly=true to surface only the contested ones.
GET /api/resolutions/{conditionId} — one market's full label plus the per-source evidence trail (on-chain payout, platform API, price-implied), each joined to that source's measured reliability.

# recent contested resolutions — the settlement-risk watchlist
curl -H "Authorization: Bearer YOUR_KEY" \
  "https://api.idmon.io/api/resolutions?reviewOnly=true"

# one market's full truth label + evidence trail
curl -H "Authorization: Bearer YOUR_KEY" \
  "https://api.idmon.io/api/resolutions/0xCONDITION_ID"

The single-market response includes labelPosterior (calibrated confidence), settlementState (finalized on-chain vs provisional), disputeFlag, and an evidence array showing what each source read and how reliable that source has historically been. That's enough to treat resolution as a measured quantity rather than a coin flip you find out about after the fact.

Try it free, no signup. Idmon's /public/* endpoints serve much of the data at 60 requests/min with no key. See the docs for what's available without authentication. For on-chain-anchored resolution truth, settlement-risk flags, real-time push, and a commercial licence, request a key — you'll be live within a business day.

What this is not

This is not financial or legal advice. Resolution rules and dispute mechanics vary by market and by jurisdiction; always read the actual contract terms before trading.
A calibrated confidence is a measured probability, not a promise. High confidence means the sources agree and history says that's usually right — not that a dispute is impossible.
Idmon reads public on-chain and platform data. It does not adjudicate disputes or influence how any market resolves.

FAQ

How does Polymarket decide the outcome of a market?

Through UMA's optimistic oracle. A proposer asserts the outcome with a bond; if nobody disputes it within a challenge window, it stands and the Conditional Token Framework contracts on Polygon pay out the winning side. If it's disputed, UMA token holders vote. The final payout is recorded on-chain.

Can a Polymarket market resolve unexpectedly?

Yes. Ambiguous wording, unanticipated edge cases, timing details, and disputes can all push a resolution somewhere traders didn't expect from the price. The price reflects probability, not certainty about how the rules will be applied.

Do Polymarket and Kalshi resolve the same event the same way?

Not always. Kalshi settles against its own published rules as a regulated exchange; Polymarket settles via the UMA oracle on-chain. Two markets that look like the same question can carry different wording and sources, so they can settle differently — that's settlement-mismatch risk.

What is settlement risk in prediction markets?

The chance a market resolves differently from what the price and the obvious reading imply — because of ambiguous terms, a dispute, an edge case, or a cross-venue rule difference. It's distinct from being wrong about the event: you can call the event correctly and still lose on resolution.

How do I get prediction-market resolution data by API?

Idmon exposes GET /api/resolutions for recent resolutions with a calibrated, on-chain-anchored truth label and dispute flags, and GET /api/resolutions/{conditionId} for one market's full label plus its per-source evidence trail. See the docs for the response shape.