TRACK

… picks graded. Model calibrates after every settled result.

Failed to load data.

—

30D UNITS

1 unit = 1% of bankroll

—

WIN RATE

—

CLV HIT

beat the closing line

—

ROI

Trailing performance · current capability vs lifetime

Recursive learning means the recent window is what matters most for tomorrow's bet sizing.

Recursive learning means the recent window matters most for tomorrow's bet sizing. Drag floors to see where edge lives now.

FILTERED ( picks)

No graded picks in this window yet.

If we had only taken picks above…

SPORT tap to scope every number on this card to a single sport

CONFIDENCE FLOOR

EDGE FLOOR

PICKS

WIN RATE

ROI

UNITS

MODELS IN FORM

Model is in early calibration — picks grading now.

CONFIRMED EDGE — WORTH ACTING ON

Segments where the lower 95% confidence bound on ROI is positive after at least 100 graded picks AND out-of-sample data agrees.

Segment	Picks	Win Rate	ROI	95% CI	Units

EDGE CALIBRATION

When we claim X% edge, do those picks return X% over time?

Points on the diagonal are perfectly calibrated. Bars show 95% confidence range — wider bar = smaller sample.

MODELS UNDER CONSTRUCTION

Segments where the model is bleeding — n ≥ 30 with negative realized ROI. Each one paired with the work shipped to fix it.

Last 14 days: model commits across all segments. The recursive learning loop is running.

Recent fixes shipped

PRO TIER

Unlock the full model breakdown

Sport-by-sport and market-by-market performance, edge development signals, tier tuning status, and model changelog. Available on PRO.