Skip to content
Browse › Brier Skill Scores
Descriptive only — a measured pattern, no edge claimed.
Analytics module · as of 2026-07-25

Against the market baseline, the skill score is negative everywhere

mlb: BSS(model vs market) in [-0.464, -0.079] -- model does not beat the market; BSS(model vs climatology) in [+0.011, +0.143] (BSS>0 means it beats that baseline). soccer_intl: BSS(model vs market) in [-1.079, -0.470] -- model does not ...
View full size ↗
confirmednull (a finding)not testabledescriptivepending
Brier Skill Scores
Chart: Brier Skill Scores -- mlb: BSS(model vs market) in [-0.464, -0.079] -- model does not beat the market; BSS(model vs climatology) in [+0.011, +0.143] (BSS>0 means it beats that baseline). soccer_intl: BSS(model vs market) in [-1.079, -0.470] -- model does not ...
scripts/platformkit/analytics_showcase/out/brier_skill_scores.json2026-07-25

What it means

Brier Skill Score is 1 minus Brier over a reference Brier; positive means you beat the reference. Against climatology (a constant base-rate guess) the model scores positive (MLB +0.042), as any real in-game model should. Against the market it is negative across all sports and checkpoints -- the market's Brier is lower. This is reported plainly, not hidden.

Caveats & confounds

The most negative MLB cell is late(inn7+) at -0.4635; late-game is exactly where market prices sharpen on information the model cannot see.

Method. standard BSS = 1 - Brier/Brier_ref; refs = climatology (const sport base rate) and market

Ask Scout about this
In-game, does the win-probability model beat the market's win probability?What is a Brier score and what is the model's?Does the model beat a simple base-rate baseline?Ask anything →