Human verification
Super Match Verification
Prediction Market Radar checks published Super Match relationships through structured human review. Reviewers compare the underlying market questions, outcomes, timing and settlement terms using evidence frozen when each case is selected.
Published review evidenceStatistical reviews457 Approved451 Rejected6 Estimated publication precision98.9% Approx. 95% confidence interval97.3–99.6%
Verification results
The precision estimate uses the stored probability-sampling design weights. The displayed confidence interval uses the corresponding effective sample size as a transparent approximation. Skipped cases do not enter the precision denominator. Historical Uncertain decisions from the earlier review workflow are treated as Approved for reporting.
What is being measured?
Verification measures the precision of published Super Matches: whether markets shown together represent compatible underlying questions and outcomes. Approval concerns the relationship between the markets; it does not predict the outcome of a market or guarantee that prices will remain unchanged.
How cases are selected
Public statistical cases are selected from published Super Match relationship compositions using a reproducible probability-based process. Once a particular relationship composition has received a founder decision, later price movement alone does not create a new verification case. A meaningful change in membership, such as an additional source market joining the relationship, can make the changed composition eligible for review again.
What reviewers assess
Review focuses on proposition meaning, outcome identity, relevant timing, thresholds or rankings where applicable, and settlement compatibility. This is deliberately stricter than matching titles or keywords alone.
Frozen evidence
Each review case preserves the publication and source evidence available when it was selected. Later changes to prices or source markets therefore do not rewrite the evidence that supported the recorded review decision.
Independent diligence
Authorised investors or platform partners may be given time-limited access to a separate 50-case blinded diligence set. That investor diligence sample is independent of the public statistical sample; its assessments are recorded separately and do not alter Prediction Market Radar's published founder-verification statistics.