WeBall.

The model

WeBall grades every pregame bet the same way: build a fair probability from recent samples and cross-book consensus, compare it to the best available price, and publish the gap.

How a grade is made

  1. 01Sample. Pull the recent game results behind the market — points, goals, yards, whatever the line is about.
  2. 02Model probability. Fit the sample and price the line into a win probability for the bet.
  3. 03Fair price. Strip the vig from the cross-book consensus to get the market's true number.
  4. 04Edge. Model probability minus the implied probability of the best price — that gap is the number on every badge: +5.4%

What the verdicts mean

GOOD

Model probability clearly beats the best available price.

LEAN GOOD

Positive edge, but modest — directional, not a hammer.

NEUTRAL

Market and model agree. No bet.

LEAN BAD

Slightly overpriced — the other side is closer to fair.

BAD

Clearly overpriced against the model's number.

THIN DATA

Not enough sample to trust the number. Informational only.

Reading confidence

Confidence measures how much data sits behind a grade — sample depth, book coverage, and consistency. It changes what you should do, not just how the bar looks.

High0.78

Model has deep, consistent data behind this number.

Medium0.45

Decent sample — treat the edge as directional.

Low0.18

Thin data — informational only, size down or pass.

Calibration ledger

Building

Every graded bet is logged when it settles. Once enough have settled, this section shows the record straight: predicted win rate vs actual by decile, cumulative ROI at the graded price, and sample sizes — wins and losses alike. Until then, no track-record claims. A confidence number you can't audit is decoration.