The Mammoth are the model's lean over the Penguins on the underlying numbers

The model is leaning Utah, and the case is about how the Mammoth have been playing, not how their results look. Utah has been the better team at the part of hockey that produces goals, and that is the edge the model is betting on. Goaltending is the one place where Utah looks exposed, which makes this a lean rather than a lock.

The model gives Utah a 57% win probability. Most of that rests on expected goals, the model's estimate of how likely each shot was to become a goal. Utah has taken 58.4% of the expected goals at 5-on-5 this season, and Pittsburgh sits below half at 49.3%.

Utah also takes more of the shot attempts, 56.2% of them, which the model reads as territorial tilt. Expect the Mammoth to spend more of the night in the attacking zone. Form points the same way: Utah is 6-3-0 through nine games, the hotter team on points pace.

The soft spot is in goal. K. Vejmelka's save percentage was 70.6% in his last start, a game the model reads as one he was beatable in. S. Murashov has given up more goals than his shots should have allowed, at -0.6 goals saved above expected across four games. K. Vejmelka is weaker by that measure, at -2.4 over eight games, though their raw save percentages sit close together.

The model's case is the underlying work. If it goes wrong, the goaltending is where it will show first.

Data: NHL API, Backcheck xG model