Market efficiency in prediction markets
Do prediction market prices reflect what's known? The hypothesis stated precisely, the conditions it depends on, and where those conditions are absent.
Updated November 10, 2026 · Lumina Markets
On this page
People say the market knows. The underlying argument is narrower: a market price reflects the information available to the people trading it. Not that the price is correct, and not that it's the best available forecast. The claim is that public information someone has acted on with money at stake tends to show up in the price.
The idea is not specific to prediction markets. It's the same efficient-markets argument made about equities, where it has been studied for decades and where the conclusion is generally conditional rather than absolute: large, heavily-traded stocks are usually treated as difficult to beat on public information, while thinly-traded ones are treated with more caution. Event contracts inherit both the argument and the conditions attached to it.
How a priced market differs from a poll
A poll records what a sample says, and nothing follows for the respondent either way. A market records what participants were willing to transact at, and being wrong has a cost. Three differences follow:
Stating a view is free; holding one at a price is not. A participant who prices badly over a long enough run will tend to have less money committed than one who doesn't, so poorly-priced views generally carry less weight over time, though that's a tendency across many participants and many contracts, not something that resolves on any single market.
A poll is a snapshot with a field period. A quoted price can change within seconds of new information, whenever a participant acts on it.
A pollster reaches the sample they contacted. A market is open to anyone who thinks the quote is wrong, including people a survey would never reach.
None of that makes a price correct. It makes it a different kind of estimate, one that costs something to disagree with. That is roughly as far as the wisdom of crowds argument goes on its own.
What "efficient" is generally taken to mean
In this context efficiency is usually framed as a claim about difficulty rather than truth: that after costs, it is generally hard to do better than the price consistently. A market can be efficient in that sense and still be wrong on any individual question. The two statements are compatible, and conflating them is what makes the compressed version misleading.
The equities parallel is useful here too. Saying a large-cap stock is efficiently priced isn't a claim that its price is right; it's a claim that a given participant is unlikely to systematically identify when it isn't. The same distinction applies to a contract.
So the question that tends to be productive is not "is this price correct?" but "what conditions is this price being produced under?"
The conditions the argument depends on
The aggregation argument is generally made about markets that are:
Meaningful volume, a narrow spread, orders resting on both sides
The price has had time to absorb information and be traded against
The settlement terms are settled in advance and not open to interpretation
Enough attention that public information is likely to be acted on
A heavily traded election contract settling on certified results sits close to that description. Not every listed contract does.
Where those conditions are typically absent
Thin markets. Where volume is low, a quote may represent a small number of participants rather than an aggregate. A contract quoted 40 bid / 52 ask isn't a well-defined 46%; a spread that wide generally indicates how little has transacted. Liquidity and spread are usually the first details looked at for that reason. Thinly-traded small-cap equities are the standard analogue.
A contract listed the same day has had little opportunity to aggregate anything, whatever number it shows. Early price movement on a new market can reflect the market finding a level rather than new information about the outcome.
The argument assumes some independence among participants. Where they're drawing on a common set of sources, the same error can be reflected many times over, which can produce a narrow price without much diversity behind it. Herding, anchoring on the visible quote rather than an independent estimate, tends to work in the same direction, since it turns prior quotes into apparent agreement. From outside the market, a narrowly-held view and a well-supported one can look similar.
Where settlement wording is unclear, part of the price can reflect disagreement about the question rather than the outcome. An awards contract can price a category dispute; a climate contract can price which dataset applies. That is genuine disagreement, but it isn't a probability about the world.
A contract resolving years out ties up money that has other uses, and there is less immediate reason for participants to do the analytical work now. Prices far from resolution are generally treated as carrying less information than near-dated ones, for the same reason long-horizon forecasts of anything usually are.
Which market details these conditions correspond to
The conditions above map onto details a market page displays:
| Detail | Corresponds to |
|---|---|
| Spread | How narrow the quote is |
| Volume | How much has actually transacted, and how recently |
| Time since listing | How long the contract has been trading |
| Price history | Whether movement lines up with public events |
| Settlement terms | Whether the resolution source is named specifically |
How to read a prediction market price covers these in the order they appear on a market page.
What the argument does not establish
Where the conditions above hold, the price is an estimate produced by participants with money at stake. Where they don't (a thin, newly listed or ambiguously worded contract), it generally reflects less. Neither statement says anything about whether a given price can be improved on, and the second isn't the inverse of the first: the same thinness that makes a quote uninformative also tends to widen the spread and make execution less predictable. Inefficiency is not a synonym for advantage.
An apparent mispricing can persist, widen, or prove to be an error in the reader's own estimate, and a binary contract that resolves against you pays zero. Nothing above identifies an advantage, suggests one is available, or recommends any position.
Lumina Markets Inc. is a CFTC-registered Introducing Broker and Member NFA (NFA ID: 0576536). Trading event contracts involves substantial risk of loss and is not appropriate for all.
FAQ
Are prediction markets efficient?
It's a hypothesis rather than a settled property, and the conditions it depends on (liquidity, time on the board, unambiguous settlement, sustained attention) are present in some markets and absent in others. It's generally assessed market by market rather than asserted about the category, in the same way the question is handled for equities.
Does market efficiency mean the price is correct?
No. Efficiency is usually framed as a claim about difficulty, that it's hard to do better than the price consistently after costs, rather than a claim that the price is right. A market can be hard to beat and still be wrong on a given question.
Is the wisdom of crowds real?
The mechanism it describes is real: participants transact at a price and bear a cost for being wrong, which a poll respondent doesn't. Whether that produces a good estimate on any particular market depends on how much independent judgment and volume sit behind the quote, which varies market to market.
What details describe how a price was produced?
The spread, the volume and how recent it is, how long the contract has been listed, whether price movement lines up with public events, and whether the settlement terms name a specific source. These are descriptive details about how a market is trading, not a checklist and not a basis for concluding anything about whether a price should be acted on.