Free during beta —Free during beta — to track favorite players + get game alertsto track favorites + alerts
MLB·Model edges
Where the model
disagrees with Vegas.
Tonight's slate ranked by the gap between the model's home-team win probability and the moneyline's de-juiced implied probability. The list runs largest gap first. Edges under 3 percentage points sit inside model noise; sharps tend to live at five or higher on the same side.
Slate
6 games
With odds
3
Top edge
+8.5pp
MLB · model edges · tonight
Where the model disagrees with the price.
6 games priced · 3 of 3 shown
EV filter
Pick: STL+12.2%EV
STLvsKC
Model56.4%
Market47.9%
Best price-101 · ML
Pick: MIL+11.8%EV
MILvsCLE
Model66.9%
Market57.2%
Best price-149 · ML
Pick: NYY+8.4%EV
NYYvsCHW
Model66.2%
Market58.4%
Best price-157 · ML
Magnitude is the percentage-point gap between our model's home win probability and the moneyline's implied probability. Edges under ~3pp are at the edge of model noise; meaningful sharps tend to live at 5+ on the same side.
Calibration · the receipt nobody else shows
Are the model's probabilities honest.
When the model says 60%, does the team actually win 60% of the time? Each row plots predicted probability against observed win rate across 1 league · 1,107 games · last 90 days. Closer to the dashed line is better.
-1.88pp
avg ECE drop · raw → calibrated
MLB
1,107 games
ECE2.72%→0.84%↓ 69%
biggest gap · under-priced 30%-40% · 9.1pp
0.84%
ECE cal
ECE = expected calibration error · weighted avg gap between predicted prob and observed rate across deciles. Lower is better. Calibrated values use a Platt scaler refit on the rolling 90-day window.
21+Gambling Problem? Call 1-800-GAMBLER.
Sportsbook links are affiliate placements — we may earn a commission when you sign up. This is analysis and research, not a guaranteed outcome. See Responsible Gambling · Betting Terms.