A $7 million Polymarket contract on the Ukraine mineral deal resolved Yes in March 2025 — not because the deal happened, but because a single UMA whale controlling 5 million tokens voted it through. That is the least glamorous corner of prediction markets, and the most important. Resolution is where your paper P&L becomes real money, or does not. Understanding it is not optional.
This piece walks through what actually happens between market close and payout on Oddup, how our approach compares with Kalshi and Polymarket, and why the Maybe leg cuts a whole class of resolution edge cases that plague two-outcome books.
The three parts of resolution
Every prediction market resolution has three moving parts. Get any of them wrong and traders lose confidence, or worse, lose money on a market that never had a clean outcome to begin with.
- The data source. Where does the market look for the answer? A named agency (BLS, Fed, CFTC), a specific news outlet, an exchange price at a specific timestamp, or something looser?
- The trigger. When does resolution kick in? An event happening, a date passing, a price threshold hitting?
- The dispute window. If someone disagrees with the outcome, what happens? Who decides? How long does it take? What can they overturn?
All three matter. All three fail in different ways.
Kalshi's approach: regulated, fast, human-in-the-loop
Kalshi's markets are CFTC-designated contract-market products. Every market is drafted with a specific data source and settlement rule before it goes live. From Kalshi's own docs: "Markets typically settle shortly after expiration, but timing can vary based on market type, data source availability, and manual review requirements."
In practice that means most Kalshi markets settle within about 3 hours of the underlying event. When they don't, it is usually because the market is waiting on official data from a source agency — a Bureau of Labor Statistics print, an FOMC statement, a specific court filing.
Yes contract holders receive $1 per contract. No contract holders receive $1 per contract. Losing tokens go to zero. Settlement fees are zero for simple yes/no determinations but may apply for sub-cent scalar settlement.
The strength of this model is oversight. The Third Circuit ruled in April 2026 that the CFTC holds sole authority over Kalshi's sports-event contracts, resolving a challenge from New Jersey. The federal supervisory layer is the settlement safeguard.
The weakness is scope. Regulated markets do not cover every question a trader wants to ask, and the CFTC review process is slow. Kalshi cannot list "Will X company be acquired by Q3?" the way a decentralised market can.
Polymarket's approach: UMA Optimistic Oracle
Polymarket uses the UMA Optimistic Oracle for resolution. Anyone can propose an outcome. Anyone can dispute it. If disputed, UMA token holders vote to decide the winner. That is decentralisation working as designed.
It is also where the trouble starts. In March 2025 a whale used 5 million UMA tokens — roughly 25% of UMA voting power — to force the Yes outcome on the $7M Ukraine mineral deal market, despite the actual event never occurring in the specified timeframe.
That was not a bug. That was the governance system working as coded, with an outcome nobody outside the whale wanted. UMA's response was to introduce further safeguards, but the underlying design — settlement by token vote — remains.
The problem repeated in a different form this year. A $60M Polymarket contract on whether Strategy sold any Bitcoin by 31 May 2026 has been disputed twice and is now in front of UMA tokenholders. The market's payout hinges not on Strategy's actual behaviour, but on how a shifting token-holder base interprets the resolution criteria weeks after the event.
A cottage industry has grown around this. Guides for retail traders now recommend treating any large Polymarket contract as carrying real dispute risk. Some traders build entire strategies around detecting UMA proposals and trading the price gap between initial resolution and final vote.
None of that is a healthy resolution mechanism. It is a resolution mechanism that has become a market of its own.
How Oddup markets resolve
Oddup's approach borrows the best of both models and cuts the worst of each.
Data source is fixed at listing. Every Oddup market names its exact source and timestamp before trading opens. A market on "Bitcoin closes August above $70,000" cites the specific exchange, the specific price feed, the specific timestamp in UTC. No ambiguity at close.
Trigger is deterministic. When the timestamp passes and the source publishes, resolution runs. No human waiting on manual review for standard cases. The named source publishes the number, the smart contract reads it, positions settle.
Dispute window is bounded. If there is genuine ambiguity — the named source publishes something unexpected, the timestamp lands on a data outage, the outcome does not map cleanly to the market rules — the market flags for review within a fixed window (typically 48 hours after nominal close). During that window trading is halted and the ops team reviews against the pre-published rules. If the rules cover the case, the ops call is announced. If they do not, the market resolves Maybe — the middle band.
That last point is the design lever. Where a two-outcome market has to force every ambiguous case into Yes or No, an Oddup three-outcome market has a native pressure valve. If the outcome truly does not resolve cleanly to Yes or No, Maybe pays.
Where the Maybe leg saves resolution
Three concrete examples where the three-outcome design shortens the resolution debate:
Election night calls. A market on "Candidate X wins the popular vote" resolves cleanly if AP calls it by a defined timestamp. If AP has not called it — as happened for four days in November 2020 — a two-outcome market either settles arbitrarily or extends into limbo. On Oddup, the AP-uncalled scenario resolves Maybe, and both sides get their partial payout.
Corporate actions with vague timing. "Company X reports Q3 earnings above analyst consensus" is a common template. It fails badly when Company X reports pre-announced (some analysts publish revised numbers, some don't) or delays reporting (does the market wait or resolve?). On Oddup, the delayed-report scenario resolves Maybe, and traders who wanted exposure to the "clean beat" or "clean miss" scenarios still get paid on those legs.
Price-based markets near the strike. "Bitcoin closes above $70,000 on 31 August" is a coin toss if BTC finishes at $69,998. On Oddup, the market is designed with a range — Yes above $70K, No below $60K, Maybe in the middle — so the coin-toss zone is Maybe by construction. Traders pricing the middle get paid for pricing the middle. Traders wanting a directional bet still get a clean Yes or No leg.
Where oracles struggle: the 2026 examples in one table
Three real 2025-2026 resolutions worth studying, and what an Oddup-style listing would have done differently.
Ukraine mineral deal (Polymarket, March 2025, $7M). The market required a specific bilateral agreement in a specific timeframe. When the deal did not close in time, the plain reading was No. A UMA whale with 5M tokens forced Yes. An Oddup listing on the same event would have named the source (an official White House or State Department release, with a specific URL) and the trigger (the release published by a specific UTC timestamp). If the release did not exist by the timestamp, resolution reads No mechanically. No token vote in the loop.
Strategy Bitcoin sale (Polymarket, ongoing, $60M). The market asks whether Strategy sold any Bitcoin by 31 May 2026. On-chain data is public but interpretation is disputed: does an intercompany transfer count? Does a sub-entity's sale count? The market is now twice-disputed and awaiting a UMA vote. An Oddup listing would have specified the exact 10-Q line item, the exact filing date, and the exact quantitative threshold. If the 10-Q shows a sale by definition X, Yes; by definition Y, Maybe. Ambiguity is priced into Maybe, not fought over post-hoc.
Kalshi's New Jersey sports contracts (April 2026). Not a resolution failure but a jurisdiction fight. Kalshi's design — CFTC-supervised, human-reviewable — meant the challenge was legal, not oracular. That is the trade-off. Regulated markets take slower listing paths in exchange for cleaner settlement rails.
The pattern: named sources with named timestamps, plus a defined Maybe leg for the ambiguous middle, cover the cases where either regulation or oracle voting struggles.
What to check before taking a position
Every Oddup market shows five things at the top of its page. Before sizing up, check them:
- Source. Named agency, named exchange, or named URL. If a market says "official announcement" without naming a specific source, it is a Kalshi-style regulated market, not a decentralised feed.
- Timestamp. The exact UTC time resolution reads. A market that closes at 23:59:59 UTC on 31 August reads the source at that timestamp.
- Yes / No / Maybe boundaries. The exact numerical or event-based lines that define each leg. No fuzzy language.
- Dispute window. Duration in hours. Standard is 48 for market-data markets, 72 for event-based markets.
- Maybe conditions. The explicit list of scenarios that resolve Maybe. If the market does not enumerate these, it is not ready to list.
If any of the five is missing or ambiguous, don't take size. That is a market where resolution is going to be an argument, and arguments favour the party with more capital and more time — usually not retail.
Why this matters for prediction traders
Prediction markets are a promise: your P&L reflects your view of the world, cleanly and without argument. When resolution fails — when a whale can flip an outcome, when a source's data is late, when the rules do not cover the case that actually happened — the promise breaks. Trust is expensive to rebuild.
Kalshi solves this with regulation and human review. Polymarket solves it with decentralised voting, and the token economics of that voting have created a new asset class of dispute risk. Oddup solves it with a three-outcome design that gives ambiguous outcomes a native home, plus fixed sources, deterministic triggers, and bounded dispute windows.
The Maybe leg is not just a betting mechanic. It is a resolution mechanic. It is what happens when the world does not deliver a clean Yes or No — because that is what the world usually does.
Disclaimer: Oddup content is for educational purposes only. Prediction markets carry risk of loss. Nothing in this article constitutes financial advice. Check jurisdictional rules before trading real-money markets. Product mechanics described reflect current implementation and are subject to change; check the market page for the authoritative resolution rules on any specific market.
