- Public picks captured
- —
- Picks graded
- —
- Services tracked
- —
- Events in next 24 h
- —
- Last capture
- —
- none recorded yet · captures run every 4 hours
We grade what they publish.
PickAuditor captures the publicly visible picks of AI sports prediction services before kickoff, then grades each one against the official result and the closing odds. Premium tiers are not captured; the record is public-tier only. Every hit rate and ROI on this site is shown with the number of picks behind it, and nothing is ranked below 30 graded picks. How the audit works.
90-day leaderboard
Public-tier picks captured before kickoff, graded against official results. Ranked by flat-stake ROI over the last 90 days; a service is ranked once it has 30 graded public picks. Affiliate status never affects order.
No service has reached 30 graded public picks yet.
No service is being tracked yet. Captures run every 4 hours; a pick is graded once the official result is in. The first ranked row appears when a service crosses 30 graded public picks, and the leaderboard is recomputed daily after that.
Segments (sport, market) below 30 picks are never shown as ranked rows. Details and the grading rules are on the methodology page.
Most agreed events today
Ordered by the number of tracked services that published a public pick on the event. “Opinions” counts picks across all markets; times are UTC.
Claim vs audit
Every service page carries a claim-vs-audit table: the figure a service states on its own site, quoted verbatim, next to the figure we measured over its public picks, with the gap in points and a note on scope. The three largest gaps across tracked services are shown here.
No claims have been extracted yet. Claims are extracted weekly from each service’s own pages and appear here with the audited figure next to them. Until a service has 30 graded public picks in the claimed scope, the audited column reads “n too small” rather than a rate.
Each service’s claim table is on its profile under Claim check; see all services.
How a record is built
- Scope
Public tier only
We capture what a service shows to anyone. Locked, blurred or subscriber-only picks are counted as seen but never stored or graded.
- Timing
Captured before kickoff
A pick counts only if it was captured before the event started. Picks found after the start are kept as evidence but excluded from every rate.
- Grading
Official results and closing odds
Each pick is graded won, lost or push against the official result, and priced against the closing line of a reference bookmaker to compute ROI and CLV.
- Sample
n on every number, under 30 not ranked
Every rate carries the number of picks behind it. A service or segment with fewer than 30 graded picks is shown as insufficient data and never ranked.
Every service has a right of reply: a service can dispute any graded pick or figure, we answer within 7 days, and accepted disputes correct the record and are logged. All disputes and their outcomes are public at /disputes.