Who decides the outcome of a prediction market?

Not the event itself. Every platform resolves against written rules, and someone applies them. On Polymarket, anyone can propose an outcome by posting a 750 USDC.e bond, after which a two hour challenge period opens and unresolved disagreements escalate to a UMA vote. Other venues settle against a named settlement source under exchange rules. Read the rules before the event, because the title is only a label.

Why do two prediction markets on the same event show different odds?

Usually wording. A date boundary, a single word such as and instead of or, or a different settlement source turns a similar looking headline into a different contract. On Ravioli right now, two markets asking whether Bitcoin passes 100,000 dollars by 31 December 2026 sit ten points apart, at 59 and 49. Each has been traded by a handful of people, so the gap reflects thin activity more than real disagreement.

What does it mean when a market is closed but not resolved?

Closed means trading has stopped and no outcome has been recorded yet. Resolved means an outcome has been recorded and the market has been scored. The two are not interchangeable, and a closed market has decided nothing. Ravioli currently has 573 markets sitting in closed status. If you are looking at one of them, the result is still pending and no position has been settled.

Do prediction market prices track events or rulings?

In contested markets, the ruling. Guangzhe Wang studied 233 disputed Polymarket markets in Economics Letters in September 2026 and found that prices behave as forecasts of how a dispute will be adjudicated rather than as beliefs about the world. The oracle overturned the rule correct outcome in about 3.4 percent of hand verified cases. Once a dispute opens, the event is already settled and only the decision remains uncertain.

Does any of this apply to Ravioli, which is free to play?

The resolution questions do. Wording, settlement sources and date boundaries matter the same way whether or not money is involved. What does not apply is anything about deposits, withdrawals, fees or state legality, because Ravioli is free to play. You trade with Sense, which is free, and the prizes are real.

Prediction market resolution: who decides what actually happened

By Reva Verma, Founder, Ravioli · Updated September 17, 2026

The same question keeps surfacing in prediction market communities, usually after someone's position settles in a way they did not expect. It sounds like a complaint but it is a real structural question: the thing happened, so why did the market settle the other way? Or the reverse, why did it settle on something that did not feel like it happened at all?

The answer is uncomfortable and worth understanding before you trade anywhere. A market does not resolve because an event occurred. It resolves because a person or a procedure read a set of rules and made a call.

The title is not the contract

Adam Niedbalski put the point directly in a piece published on 13 September 2026: prediction markets do not have a counterparty problem, they have a resolution problem. Collateral handles solvency. What it cannot handle is whether everyone accepts the decision about who was right. His framing is the one to keep: market titles are not contracts, the rules are. A trader can be right about the world and still wrong about the settlement.

There is a well documented example of the gap. On 9 July 2025, after Linda Yaccarino announced she was leaving X, Kalshi changed a market's title from "Linda Yaccarino leaves X this year?" to "Linda Yaccarino replaced as X CEO this year?" As Dustin Gouker reported at the time, the underlying rules had always required that a successor be named, not merely that Yaccarino depart. By his account the market had run up to around 99 before the clarification and fell back to the 65 to 74 range afterwards. Nothing about the world changed in that window. The reading of the rule was clarified, and that was enough.

Who actually makes the call

The mechanism differs by platform, and it is worth knowing which one you are dealing with.

Polymarket routes resolution through the UMA optimistic oracle. Anyone can propose an outcome by posting a 750 USDC.e bond, and once proposed the market enters a two hour challenge period during which the outcome can be disputed. A correct proposal returns the bond plus a reward. An incorrect one forfeits the bond. Disputed markets escalate into UMA's own dispute process, which is a vote.

That vote is not a formality, and there is now research on what it does to prices. In a paper published in Economics Letters in September 2026, Guangzhe Wang of the Hong Kong Polytechnic University examined 233 disputed Polymarket markets, using a language model audit to work out the rule-correct winner independently of what the oracle decided. Two findings matter. The oracle overturned the audited-correct outcome in roughly 3.4 percent of hand-verified cases. And within six hours of a dispute opening, the audited-correct outcome already traded at a discount in the markets the oracle would go on to overturn.

Wang's conclusion is the sharpest summary of the whole problem available: prices in contested markets are "best read not as beliefs about the world but as forecasts of how a decentralized rule will be adjudicated." During a dispute the real event is already settled. What is still uncertain is the ruling, so that is what the price is about.

What this looks like on our own board

This is not a problem that only affects other people's platforms, so here is ours with the lid off.

Ravioli currently has 715 open markets. Among them are pairs that ask what looks like the same question and are not the same question at all. All prices below are the board's opening line, read this morning. Where a market has been traded, it has been traded by a handful of people, and the trader count is given so you can weigh it yourself.

| Market | Closes | Yes / No | Traders | | --- | --- | --- | --- | | Will Bitcoin surpass $100,000 USD by December 31, 2026? | 31 Dec 2026 | 59 / 41 | 4 | | Will Bitcoin's price exceed $100,000 by December 31, 2026? | 31 Dec 2026 | 49 / 51 | 3 | | Will Tesla merge with SpaceX and xAI by the end of 2026? | 31 Dec 2026 | 50 / 50 | 2 | | Will Tesla complete a merger with SpaceX or xAI by the end of 2026? | 31 Dec 2026 | 50 / 50 | 2 | | Will GTA VI be released before Holiday 2026? | 25 Dec 2026 | 50 / 50 | 1 | | Will 'Grand Theft Auto 6' be released on November 19, 2026? | 19 Nov 2026 | 80 / 20 | 3 |

The two Bitcoin questions are, for practical purposes, the same question, and they sit ten points apart. The two Tesla questions are genuinely different, because one needs a merger with SpaceX and xAI and the other needs a merger with either, and they sit at exactly the same number. The board has not absorbed the difference in the second pair and has produced a difference that is hard to justify in the first. That is what an early, thinly traded board looks like, and it is the honest description of it.

Two more things about our resolution state

Closed is not resolved. Ravioli currently has 573 markets sitting in closed status, which means trading has stopped and no outcome has been recorded. If you are looking at one of those, nothing has been decided yet.

And a limitation we would rather you hear from us. Once a market resolves here, the price shown on it is the settled value, not the line that was showing before the event. We checked 105 markets that resolved in the last seven days and carried a recorded yes or no outcome. In 100 of them the price now agrees with the outcome, clustered tightly at 5 and 95. That is settlement, not forecasting skill, and we will not present it as a track record. Until we can read the line a market showed before it resolved, we cannot publish an honest scorecard of our own calls, and we are not going to publish a dishonest one instead.

What to do with all this

Read the rules, not the headline. Find the settlement source before the event, not after. Check the exact date boundary, because "before Holiday 2026" and "on November 19, 2026" are not the same contract. And when two markets on the same event disagree, assume the wording differs until you have checked that it does not.

If you want the mechanics underneath all of this, we wrote them up in how market odds and resolution work. For where the law currently sits, there is our explainer on the circuit split.

Ravioli is a free-to-play prediction market. You trade with Sense, which is free, and the prizes are real.

Create a free account and browse the board.

Start playing free on Ravioli

Trade prediction markets and put your read on the world to the test. No real money, no risk. Just predict, trade, and climb the leaderboard.

Browse markets