The Avalanche stole one. They lost the expected-goals battle 3.36–4.53 and won anyway.
Utah dominated expected goals at 57%, but Colorado won 4-2 anyway—this was a theft.
The Mammoth controlled 30 shot attempts to 26, and the xG model gave them 4.53 to Colorado's 3.36. High-danger chances favored Utah 7-5. By every underlying metric, they should have won. They didn't, because S. Wedgewood was immovable.
Wedgewood stopped 28 of 30 shots (0.933 save percentage) and finished 2.53 goals above expected—a performance that rarely happens. He faced seven high-danger chances and turned away all seven. K. Vejmelka, on the other end, let in 0.64 more goals than his xG model predicted. That swing, 2.53 goals each direction, decided the game.
Utah's execution in high-danger areas collapsed. They went 0 for 7 on their clearest looks. Colorado, by contrast, was clinical in transition: they scored four goals on just 20 rush chances, a rate Utah couldn't match at 2 on 27. Martin Necas's slap shot from the top of the circles (second period, 17:17) marked the inflection point—the moment where underlying dominance couldn't stop the scoreboard.
The Mammoth earned this loss through possession and positioning. They created more rebounds, maintained better control, sustained pressure. But generating chances without converting them is just a prettier way to lose. This is precisely the game where one goalie steals a win that expected goals promised to someone else.
