How Prediction Pantheon matches markets across platforms
Overview
Comparing two venues only makes sense if both contracts really settle on the same event. This page explains how we decide that, what each label means, and where the method can be wrong.
What the labels mean
- Exact Match — automated checks support the same settlement event and outcome, subject to our confidence and safety gates. It is not a legal guarantee of identical payouts; read both source rulebooks.
- Similar Market — related wording is not sufficient evidence of an equivalent payout. Do not treat this label alone as permission to compare prices or build a hedge.
- Single Platform — only one venue currently lists this contract, so no cross-platform comparison is shown.
- Related Market — contracts about the same underlying subject that are not the same question. Shown for context only, never as a price comparison.
Comparison labels are shown on every card and market detail page. Where a comparison is not considered safe, we split the pair and show each venue on its own rather than publishing a misleading spread.
The checks that matter
- Question wording — normalised text similarity between the two venue titles.
- Outcome and action — whether both contracts pay out on the same direction of the same event.
- Resolution criteria — the source and rule the venue uses to settle.
- Event time — compare the observation window and cutoff, including time zone. Platform settlement time is a separate field and is not proof of event identity.
- Market structure — binary versus multi-outcome, and any price or threshold targets in the question.
- Liquidity and price sanity — implausible spreads and missing or stale prices are excluded rather than published.
Identity first, comparison readiness second
We separate two questions: do the contracts describe the same settlement event, and do we have adequate current evidence to compare them? Text similarity or a confidence score cannot override a conflicting threshold, action, date, outcome or resolution rule. Ambiguous evidence fails closed rather than becoming a claimed opportunity.
- Establish the subject, event, outcome direction and market structure.
- Check numerical thresholds, election phase, observation window, time zone and available settlement-source evidence.
- Reject conflicts or unknown critical facts. An expected event date and a platform payout date are not interchangeable.
- Only compare different platforms with supported source buy quotes and valid lifecycle/freshness evidence.
- Confirm rules, order-book depth and fees on both platforms before any real trade.
Reviewed benchmark examples — not live market claims
The following cases come from PP's frozen matching benchmark reviewed on 20 August 2026. They illustrate reviewed identity decisions, not current listings, prices or verified trading opportunities. Normalised benchmark rule excerpts are not verbatim platform rulebooks.
| Market family | Reviewed decision | Reason |
|---|---|---|
| Next French presidential election winner / Next French Presidential Election | Equivalent event in the reviewed benchmark | Same next 2027 French presidential election winner; current quote readiness was not established by this benchmark. |
| NHL Eastern Conference Finals Winner / NHL: 2027 Champion | Not equivalent | Conference final winner is not the league champion. A team can win one and lose the other. |
| 2028 US presidential winner wording / next inaugurated president rules | Related, not comparable | Election winner and next inaugurated president can diverge; the reviewed rule definitions differ despite similar titles. |
Buy prices, freshness and missing evidence
A buy-price comparison uses supported source buy quotes, not a last trade, midpoint or invented zero. On Kalshi, Yes uses the Yes ask and No uses one minus the Yes bid. If a required quote is unavailable, we do not manufacture a comparison.
PP classifies effective age as LIVE at 5 minutes or less, DELAYED above 5 and up to 10 minutes, and STALE above 10 minutes. Comparisons use the older required-leg quote and the snapshot age, whichever is older; fetching a page does not refresh the underlying source observation.
A 24-hour move requires an observed baseline within two hours of the point 24 hours earlier. Missing evidence stays unavailable, not zero. Movement is a probability difference; 24-hour volume is separate from cumulative volume.
Where matching can be wrong
Automated matching is a starting point, not a guarantee. Two contracts can share almost identical wording and still settle differently — on a different data source, a different cut-off time, or a different treatment of an edge case. A visible price difference may therefore reflect a genuine disagreement between traders, a difference in the contracts, or simply stale data on one side.
Before acting on any gap, open both contracts on their own venue and read the settlement rules, deadline, fees and available liquidity. A price gap is a prompt to verify, not proof of a risk-free opportunity.
If you find a match you believe is wrong, please report it — see our corrections policy.
Frequently asked questions
What does Exact Match actually guarantee?
It records an automated identity judgement subject to safety gates, not a guarantee of settlement or profit. Comparison readiness additionally depends on source quotes, lifecycle and freshness. Always verify both source rulebooks.
Why do you sometimes show one platform instead of a comparison?
When a pair fails our safety checks — mismatched deadlines, missing prices, or differing thresholds — we split it and show each venue separately rather than publish a false spread.
Is a price gap an arbitrage opportunity?
Not by itself. Fees, spreads, liquidity, funding costs and rule differences can erase or reverse an apparent edge. See our guide on arbitrage risks.