One full-rate play and two leans clear the model's bar tonight, led by a Montreal-Dallas total that the market is pricing well under where the model lands.
Eleven games on the board, three plays cleared the gate, and the best of them is Montreal at Dallas going Over 6.5 at +114. The model has this number at 53.9%, while the price only needs 46.7% to break even. That gap is the whole story: 15.3% expected value is not a marginal lean, it's the kind of number that should jump off the page on a Thursday slate.
The factors behind it pull from a few directions. Home injuries sit at -0.58, which hurts Dallas. Away injuries come in at -0.36, hurting Montreal. Away special teams at -0.175 is another drag on Montreal, while home special teams at +0.106 gives Dallas a boost. Mixed bag on both sides, but none of that changes the total itself — these are inputs into a number, not a read on who wins or how the game gets played.
Two caveats worth saying out loud because the model says them: no goalie is confirmed for this one, and there's no observed GSAx for either netminder. That's a real hole in the profile. A total play with no goalie information baked in is still a bet on scoring environment, not on who's actually in net. Worth sizing accordingly.
Two more clear the bar, both tagged LEAN rather than the top tier, and both move at shorter prices that eat into the margin. Boston at St. Louis Over 5.5 sits at -130, with the model at 58.3% against a required 56.5%. That's 3.1% EV — real, but thin enough that it lives or dies on getting the number rather than chasing it after the line moves.
Columbus at Carolina Over 5.5 is nearly identical in shape: -135, model at 59.0%, breakeven at 57.4%, for 2.8% EV. Both leans clear the threshold by a point or two and nothing more. These aren't spots to load up on. They're spots where the math works, barely, and the juice on each side limits how much room there is for error.
Every EV figure here is measured against the actual price on the board, not some vig-stripped fair line. That distinction is the whole ballgame. A total can sit a fraction off what the model considers fair and still lose money once the book's margin — typically 2 to 2.5 points on an NHL line — is baked in. The Dallas-Montreal number clears that bar with room to spare. The two leans clear it, but just barely, which is why they're graded differently.
This is one model's output on a small sample of a single day's slate. No track record is being claimed here, and none should be assumed. Treat it as analysis, not instruction — and treat the goalie uncertainty on the top play as a real unknown, not a footnote.