On this page (7 sections)
Content reviewed: 2026-10-13. Examples are hypothetical.
A market is potentially mispriced for your purposes when your evidence-based probability estimate differs from the cost of obtaining the corresponding payout. The conclusion depends on your estimate being useful; disagreement with the crowd alone is not an edge.
Define the question and your baseline
Read the resolution criteria, then write the probability of that exact outcome before using the quoted price as an anchor.[3] Start with a relevant reference class if one exists. Explain why the current case should differ, using evidence available before the trade.
Keep an uncertainty interval alongside the point estimate. If your plausible range is 45%-65%, calling a 54-cent contract an obvious bargain because the midpoint is 55% overstates the evidence.
Compare against the all-in purchase cost
For a $1/$0 contract, the expected profit per unit is p - c, with entry fees included in cost c. A hypothetical 60% probability and a 52-cent all-in cost give an eight-cent expected profit. If your probability estimate falls to 50%, that becomes a two-cent expected loss.
The displayed mark can differ from the executable ask, and larger purchases may reach more expensive levels.[1] Use the order-book guide to estimate the cost at your intended size. If your strategy depends on selling before settlement, model that exit separately rather than substituting a $1 payout.
Ask what the other side may know
List alternative explanations for the gap: a different interpretation of the rules, stale evidence, an upcoming revision, transaction costs or a trader with better information. Try to invalidate your thesis before increasing the size. A market can stay away from your estimate longer than your capital is available.
Test the forecasting process
Save forecasts with timestamps, then score resolved outcomes. Proper scoring rules evaluate probabilities rather than merely counting correct binary calls.[2] For the binary Brier loss, use (p - outcome)^2, where outcome is 1 or 0. A 60% forecast scores 0.16 on a win and 0.36 on a loss. Compare an average across a defined sample against a simple baseline forecast made at the same times.
Do not select only successful trades for the record. Separate model quality from execution quality: a useful forecast can be traded badly, and one profitable result can arise from an overconfident forecast.
Decide whether the margin is worth using
Use a smaller position or no position when reasonable changes to your estimate eliminate the edge. The Kelly guide explains a sizing model, but its output is only as reliable as the probability and payoff inputs. A spreadsheet's precision does not reduce uncertainty about the event.
Try the calculation
Try the expected value calculator with its starting example: a 60% estimate, a 50-cent price and a $100 budget. Before fees, the expected profit is $20; it is an average under your estimate, not the profit from either individual outcome. Add the order fee and inspect the probability sensitivity table.
References
- Polymarket Documentation. Prices & Orderbook. Accessed October 13, 2026.
- Tilmann Gneiting and Adrian E. Raftery. Strictly Proper Scoring Rules, Prediction, and Estimation. Journal of the American Statistical Association 102(477), 359-378, March 2007. DOI: 10.1198/016214506000001437.
- Polymarket Documentation. Resolution. Accessed October 13, 2026.