A prediction market is mispriced only when the price you can actually trade differs meaningfully from a defensible probability estimate. A dramatic headline, a model disagreement, or a whale trade can point you toward a market, but none of those signals proves that an edge still exists.
The useful workflow is simple: define the contract, estimate the probability, inspect the executable price, subtract trading friction, and decide what evidence would invalidate the thesis. This guide turns that process into a repeatable checklist for Polymarket, Kalshi, and other event-contract markets.
The short version: price is not the same as your fill
A market displayed at 42% may not be buyable at 42 cents. On an order book, buyers post bids and sellers post asks. If the best bid is 39 cents and the best ask is 45 cents, the displayed midpoint may be 42 cents while an immediate buyer pays 45 cents.
Polymarket's official prices and order-book documentation makes the distinction explicit: the displayed price can be the midpoint, but an immediate buy executes at the ask and an immediate sale at the bid. Kalshi likewise exposes resting bids and asks in its order book.
Your edge must survive the spread, fees, and slippage. A model estimate of 46% against a 42% midpoint is not a four-point edge if the position costs 45 cents to enter.
A six-step mispricing checklist
1. Rewrite the market as an exact test
Before estimating anything, write down the precise condition that pays $1. Include the deadline, source, geographic scope, and edge cases. A contract asking whether an announcement happens by Friday is different from one asking whether a policy takes effect by Friday.
Read the full resolution rules rather than relying on the title. Polymarket's resolution guide notes that the rules define the source, end date, and treatment of ambiguous cases. A good thesis on the real-world event can still lose if it answers a different question from the contract.
2. Build an outside-view baseline
Start with a base rate before reading today's most persuasive argument. Useful anchors include historical frequencies, comparable events, polling averages, option-implied expectations, official schedules, and forecasts made before the latest news cycle.
The outside view keeps one vivid headline from becoming the entire model. If similar events occurred 18% of the time, moving immediately to 70% should require unusually strong evidence.
3. Add contract-specific evidence
Now update the baseline with information that directly changes the contract's outcome:
- Freshness: when was the evidence published or observed?
- Source quality: is it an official filing, primary data, credible reporting, or social speculation?
- Independence: are five articles repeating one unnamed source?
- Direction and magnitude: should the evidence move the probability two points or twenty?
- Contract fit: does it affect the exact deadline and resolution test?
Use Alphascope's prediction-market news to map catalysts to affected contracts, then open the original source before changing your estimate.
4. Compare with related markets
Related contracts are a powerful consistency check. A presidential nominee market should make sense beside state-primary odds, endorsement markets, and the general-election market. A Fed decision contract should be coherent with inflation, jobs, and later-meeting probabilities.
Cross-platform prices can also expose a disagreement, but only when the contracts truly match. Compare deadlines, settlement sources, cancellation rules, and payout definitions before treating a Polymarket-Kalshi gap as evidence.
5. Inspect the executable order book
Record the best bid, best ask, available size, and spread. Then estimate the average fill for the amount you intend to trade. Polymarket's official order-book tooling includes depth and estimated fill calculations because the top-line price alone is not enough for position sizing.
| Input | Example | Why it matters |
|---|---|---|
| Your probability | 58% | Your evidence-based estimate |
| Displayed midpoint | 52% | Useful reference, not necessarily tradable |
| Best ask | 55 cents | Immediate entry price |
| Estimated average fill | 56 cents | Accounts for your intended size |
| Pre-fee edge | 2 points | 58% estimate minus 56-cent fill |
In this example, the apparent six-point midpoint gap shrinks to two points before fees and estimation error. That may be too small to justify a trade.
6. Set an uncertainty margin and invalidation rule
A probability estimate is not exact. Write a range, such as 54% to 62%, instead of pretending 58% is certain. If the executable price sits inside your reasonable range, the market may be fairly priced even when the point estimate differs.
Also state what would make you change your mind: a new poll, court ruling, injury update, official data release, or a change in the resolution source. This turns monitoring into a process instead of an emotional reaction to price movement.
What a useful Polymarket scanner should surface
A scanner should narrow the research queue, not issue blind picks. The most useful outputs are:
- the exact contract and current bid-ask spread;
- available liquidity at realistic position sizes;
- the catalyst and its publication time;
- related or equivalent markets;
- an independent probability estimate with assumptions;
- resolution-rule warnings;
- the difference between midpoint and executable price.
Use the Alphascope Polymarket analytics and AI analyzer to structure that review, then verify every important input against the live market.
Five false positives that look like an edge
- Stale last trade: the displayed price came from an old transaction while the current order book moved.
- Wide spread: the midpoint looks attractive but the ask removes the gap.
- Mismatched contracts: two platforms use different deadlines or resolution sources.
- Headline duplication: many posts repeat one fact the market already incorporated.
- Thin liquidity: a small order moves the price through several levels.
A pre-trade worksheet
- Contract: What exactly pays $1?
- Baseline: What was the outside-view probability before today's news?
- Update: Which evidence changed the estimate, and by how much?
- Range: What is the reasonable low-high probability interval?
- Execution: What are the bid, ask, spread, depth, fees, and expected fill?
- Correlation: Which other open positions depend on the same outcome?
- Invalidation: What new information would close or reverse the position?
The goal is not to trade every disagreement. It is to reject weak signals quickly and spend attention where the probability gap remains after all the hard checks.
FAQ
What is a mispriced prediction market?
It is a market where a defensible probability estimate differs enough from the executable price to survive spread, fees, slippage, uncertainty, and resolution risk.
Does a whale trade prove that a market is mispriced?
No. The wallet may have a different entry, hedge, time horizon, bankroll, or information set. Treat wallet activity as a research trigger, not a complete thesis.
How large should a prediction-market edge be?
There is no universal threshold. The required margin should grow with uncertainty, spread, fees, thin liquidity, ambiguous rules, and correlated exposure.
Can AI find mispriced Polymarket contracts?
AI can scan news, compare markets, and create probability estimates, but it can also use stale evidence or misread rules. Verify the contract, sources, and executable price manually.