Data refreshes every morning around 8:00am ET when the model publishes new outputs.
The edge exists because the market is pricing Dallas as a multiple-goal favorite while the model sees the Stars winning by less than one on average. That gap is meaningful on a puck line, where the difference between favorite control and actual cover rate is substantial. Utah only needs to keep the game competitive, and the projection suggests they do that often enough to justify a play. Even if Dallas dictates territory, the underdog side is protected by the frequency of one-goal finals and by the volatility of NHL finishing.
This edge comes from a classic favorite-versus-puck-line gap: Colorado can be the superior team and still fail to cover by two. The model projection of roughly Avalanche by a goal makes Pittsburgh +1.5 attractive, especially in a matchup where one-goal outcomes remain common. Pittsburgh's spread value is amplified if they can generate enough offense to avoid being chased out early and force Colorado to protect rather than extend. With no meaningful total edge, the cleaner angle is simply taking the cushion on the dog.
The edge exists because the market asks Detroit to win by multiple goals while the model projects a much tighter game. Calgary +1.5 benefits from a lower-scoring game script and from the inherent variance of NHL one-goal finishes. Detroit may still be the better side outright, but laying -1.5 requires cleaner finishing and empty-net margin that the projection does not fully support. In a game priced close enough by the model, taking the underdog cushion is the more efficient spread position.