Trading Ambiguity: How to Price an Unclear Resolution Rule on Polymarket
A practical method for separating probability risk from rule risk, estimating an ambiguity premium, and sizing a No position when a contract can resolve 50/50.

Ambiguity is not the same as uncertainty. Uncertainty is not knowing whether the event will happen. Ambiguity is not knowing what the contract will pay if it does. A market can price the first precisely and the second badly, and the gap between them is where a trader can be right about the world and still lose money.
Every contract carries two probabilities. The first is the probability of the event. The second is the probability that the rule resolves the way you read it. Most traders estimate only the first. The second is harder to model, it is correlated across markets, and it is exactly the risk that a dispute can turn into a 50/50 payout.
Two kinds of risk on one price
Decompose the contract into three states: a clear Yes, a clear No, and a contested outcome that pays 50 cents. The expected payout on the Yes token is the probability of a clear Yes plus half the probability of a contested outcome. A price of 12 cents can be fair if the clear Yes probability is 8% and the contested probability is 8%. It can also be fair if the clear Yes probability is 12% and the contested probability is zero.
Same price, different asset. Two markets at 12 cents are not the same trade if one resolves on an official government release and the other resolves on whatever a reasonable observer concludes. The first is a probability bet. The second is a probability bet plus a rule bet, and the rule bet has a wider distribution.
The ambiguity premium
Buyers of Yes in an unclear market demand a discount. Sellers demand compensation for the chance of a contest. Which side dominates depends on the direction of the ambiguity and on the crowd's reading of the rule.
For a longshot with a vague trigger, the Yes is often cheap relative to a naive reading, because the market charges for the chance of a No or a 50/50. For a favorite with the same vague trigger, the No is expensive for the same reason. That is an ambiguity premium, and it is distinct from the favorite-longshot bias. The bias is about how people weight small probabilities. The premium is about the payout distribution being wider than binary.
The premium can also be negative. If the crowd reads an ambiguous rule optimistically, the Yes can trade above the probability of a clear Yes, with the difference paid for hope rather than fear. The sign is an empirical question.
Buying No on an ambiguous longshot
Start with a clean case. A contract at 5 cents with a clear rule and a 3% true probability has a No price of 95 cents. The politics taker fee is 100 x 0.04 x 0.95 x 0.05 = $0.19 per 100 shares. Expected value per share is 0.97 - 0.95 - 0.0019 = +0.0181, or 1.9% of the 95-cent stake. Thin, but positive.
Now add ambiguity. If there is a 10% chance the market resolves 50/50, the expected payout on No is 0.87 x 1 + 0.10 x 0.5 = 0.92. Against an all-in cost of 95.19 cents, the expected value is 0.92 - 0.9519 = -0.0319 per share, a loss of 3.4% of capital. A 10% ambiguity probability turns a +1.9% trade into a -3.4% trade.
The break-even ambiguity probability is 0.0181 / 0.5 = 3.6%. The No trade only survives if the chance of a 50/50 resolution is below about 3.6%. That is a narrow window. It also explains why selling every ambiguous longshot is a bad rule. At 5 cents you are risking 95 cents to make 5, and a single void erases about ten wins.
Source and deadline risk
Source risk is the most common form of ambiguity. The rule names a source, or worse, a category such as credible reporting. A source that publishes and then retracts creates a state the rule may not cover. A source that is paywalled or inaccessible creates a verification problem. A source that reports a different number from the official record creates a conflict.
Deadline risk is the second. By Friday in which timezone. Whether the rule uses the date of the event or the date of the announcement. Whether an event that begins before the deadline and concludes after it counts. A market on an announcement can trade at 70 cents on a leak, then collapse when the deadline passes without an official statement. That is a rule trade, not a probability trade, and it can happen while the underlying event is still likely.
A useful test is to write the Yes condition as one sentence and try to break it. If it needs a second sentence to handle a postponement, the market has an ambiguity.
UMA disputes and the cost of being contested
Most Polymarket markets resolve through UMA's optimistic oracle. A proposer posts a bond, typically $750, and a two-hour challenge window follows. A dispute triggers a second proposal round. If that is disputed too, the question goes to UMA's Data Verification Mechanism, where token holders vote with a 65% majority after a 24-hour commit period and a 24-hour reveal. An undisputed market resolves in about two hours. A disputed one can take four to six days.
Disputes are rare. A 2026 reconstruction of Polymarket oracle activity counted 1,604 disputes against 184,148 request creations, well under 1%. Low base rates are exactly what makes the risk easy to underprice. It does not appear in most samples, and when it does, it is a large jump rather than a small one.
The cost is not only the outcome. Capital is locked for days, the price can gap, and the final resolution can differ from the last traded price. A position sized for a two-hour settlement is mis-sized for a six-day one. UMA describes the DVM as a Schelling point mechanism, which is a precise way of saying the outcome depends on what stakers collectively believe the right answer is. For clean questions that is usually fine. For questions that turn on interpretation, it is a second forecast layered on the first.
Probability risk versus rule risk
Probability risk is continuous, diversifiable and priced by the market. You can be wrong about an election and right about the average of ten elections. Rule risk is binary, correlated across markets that share a template, and often mispriced because reading rules is boring.
Two markets that both say official results can resolve on the same contested source. Holding both is one rule bet, not two. Five contracts that all depend on the same government release are one rule bet with five tickers. Cap the family, not the market.
Sizing follows. If a 50/50 outcome would cost half the position, ten positions in the same rule family can all be hit by one dispute. Treat rule risk as a separate budget, and cap the family at the size you would give one clean market.
How to execute
- Read the full rules, not the title, and write the Yes condition in one sentence.
- Identify the resolution source and confirm you can access it.
- Search the clarification history for additional context updates.
- Convert the deadline to UTC and note whether it is the event date or the confirmation date.
- Check the oracle status before entering. A live proposal inside a challenge window is a different risk from a market that has not been proposed.
- In an ambiguous longshot, prefer the No side as a maker. Crossing the spread at 95 cents leaves almost no room for error.
- Size for a 50/50 outcome, not for the base case.
- If you cannot hold through a six-day dispute, exit before the proposal window rather than after.
Risks
- You may be wrong about the ambiguity. The market may have read the rule correctly and your unclear contract may be clean.
- The premium can be overpriced. Paying 95 for a No whose fair value is 93 is not a trade, even if the rule is vague.
- A DVM vote is governance, not law. Outcomes can surprise.
- Rule families are correlated. Diversifying across tickers is not diversifying across templates.
- The 50/50 outcome is a real possibility, not a theoretical footnote.
Checklist
- Can I state the Yes condition in one sentence?
- Is the resolution source named, public and stable?
- Does the rule specify a date, a time and a timezone?
- What happens on postponement, withdrawal or a partial event?
- What is the probability of a 50/50 resolution, and does the price survive it?
- Is this rule template shared with other positions in my book?
- Can I hold the position through a six-day dispute?
- Is the No side still positive after the fee at the price I can actually get?
Sources
- 1.Resolution— Polymarket Documentation
- 2.How does UMA resolve prediction markets?— UMA Documentation
- 3.How does UMA work?— UMA Documentation
- 4.Resolution Is Not Settlement, Part I: Oracle Adjudication and Semantic Governance on Polymarket— arXiv
- 5.Understanding Prediction Markets and Event Contracts— U.S. Commodity Futures Trading Commission
- 6.Risk, Ambiguity, and the Savage Axioms— The Quarterly Journal of Economics