Eleven games on the board, three plays cleared the gate — but only two actually make the cut once the model's own caveats get weighed.
Eleven games projected today. Three plays cleared the threshold we actually care about: the model's side has to sit at 50% or better, and the price has to return at least 2% expected value. Both conditions, not one.
Worth saying plainly: that EV is measured against the real number you'd actually bet, not some theoretical no-vig line. A play can look fine against a fair price and still be a loser once the book's margin is baked in — call it 2 to 2.5 points on a typical NHL moneyline. The two plays below clear that bar with the actual price attached. A third did too, on paper, and we're going to explain why it's getting passed anyway.
On raw EV, this is the best number on the board. Model has Chicago at 69.3% to cover +1.5, the price at +225 only needs 30.8%, and that gap works out to 125.4% expected value. That's an enormous number for a puck line.
The factors behind it: away injuries hurt Los Angeles, home high-danger rate hurts Chicago, away special teams hurts Los Angeles, and home 5v5 xGF versus opponent xGA hurts Chicago. Mixed signals pointing different directions, which is part of the problem.
The model itself flags three issues here: no confirmed goalie for either side, no observed GSAx for either netminder, and the model is sitting well off the book's own view of who the favorite should be. When a number this big comes with that much structural uncertainty attached, it's a pass. The gate let it through. The caveats don't.
This is the top actual play on the board. Model has Columbus at 57.9% to win outright, the -120 price needs 54.5%, and that's 6.1% EV — modest, but clean. No flagged disagreement with the market here, no asterisk attached in what we were given. Smaller number than the Chicago line, more trustworthy one.
Second play clearing the gate: the under at 6.5 goals, priced at +105. Model has it at 51.4%, the number needs 48.8% to break even, landing at 5.4% EV. Plus-money on a total that the model likes slightly more than a coin flip is the kind of spot that doesn't need a big model edge to be worth the price.
All of this is one model's output on a small sample for one day's slate. It is not a demonstrated edge over time, and it isn't financial advice — treat it as what the numbers say today, nothing more.