A calibrated statistical model, priced against the real odds market, with every call published,
timestamped and graded against the closing line — in the open, win or lose.
Measurable edge vs fair odds Honest uncertainty — probabilities, not promises Real market prices, timestamped Every call logged & graded in the open
The numbers we judge ourselves on
-0.25%
avg closing-line value
across 53 graded flags
45.3%
flags that beat the close
24 of 53
8 of 20
of the last 20, beat the close
3
shadow picks graded
every scanned fixture, not just posts
Why closing-line value, not profit? CLV is the professional benchmark for real
edge — it converges in hundreds of data points, not thousands. Our live sample is young, so we make
no performance claims: we show you the process and let the ledger speak. See the full record →
Latest receipts — flagged price vs the close, win or lose
América Mineiro to Win
América Mineiro v Londrina
2.02 → 2.18
-7.3%
Both Teams to Score
Sandefjord v HamKam
1.70 → 1.65
+3%
Bodø/Glimt to Win
KFUM v Bodø/Glimt
1.57 → 1.53
+2.6%
Rosenborg to Win
Rosenborg v Kristiansund BK
1.66 → 1.62
+2.5%
What you get
Data-driven
A Dixon-Coles + ELO model rebuilt continuously from a growing archive of real results, blended with the market it is priced against.
AI-assisted, bounded
Grounded AI briefs explain the numbers and flag risks. Models never invent a price or a probability — the statistical engine and the market own the numbers.
Market-focused
Every idea is compared against the sharp line and devigged fair odds — value is measured, not asserted.
Logged & graded
Every prediction is timestamped and graded against the closing line. Wins and losses stay on the page.
Built for analysts
Signal-level transparency: see what each input contributes, its sample size, and whether it earns its keep.
Always improving
The model retrains on every finished match, and weak signals get demoted by evidence — not by opinion.