Dataset: 1,552 MLB games (2026 season through July 26), flat $100 per game
Favorites lose money everywhere — that's just vig. But division favorites lose less than non-division favorites, which is counterintuitive. The real story is on the dog side: division dogs are roughly breakeven (-0.51%) while non-division dogs are slightly profitable (+0.72%). On the surface, division games look like a wash.
The standard deviation tells a consistent story across all four segments: favorites cluster tighter (~85-86) because you're risking $100 to win less. Dogs are wider (~110-111) because the payouts are larger when they hit.
The aggregate hides a clean split. When you break division games by who is favored, two dead zones and one live edge appear.
Home favorites in division games are priced almost perfectly. 59.2% win rate, -1.19% ROI — that's the vig and nothing else. The market understands this spot well. Betting the away dog against a home division favorite is a losing proposition (-7.74%).
Away favorites in division games are where the market breaks. These teams win only 48.0% of the time — they're favorites who lose more often than they win. The -15.24% ROI is not a small leak, it's a structural mispricing. The home dog in this spot wins 52.0% at dog prices, producing +10.48% ROI across 179 games.
When the away team is favored in a division game, two forces collide:
- The away team is the better team on paper — that's why they're favored despite not having home field.
- The home team knows them intimately — same division means 13-19 meetings per season. The home team has seen this pitching staff, knows the tendencies, and plays in their own park.
The market prices the talent gap but underprices the familiarity discount. A division home dog isn't a random underdog — they've faced this opponent repeatedly, with the comfort of their own ballpark, their own bullpen matchups, and a crowd behind them. The “nothing to lose” psychology compounds it: they're already expected to lose, so they play loose.
The standard deviation supports this. At 108.01, the away-fav home-dog line has the lowest variance of any dog segment, meaning the outcomes are tighter than typical dog bets. This isn't a high-variance lottery ticket — it's a consistent edge.
This is a single-season dataset (2026, through July 26). 179 games is a reasonable sample for a game-level edge, but it's not three seasons of walk-forward validation. The signal is strong enough to shadow-track and build filters around (pitching matchup quality amplifies the edge significantly), but it should be treated as a hypothesis with promising in-sample results, not a proven system.
Hot Forum Topics
