The Sabres won a game that looked like a coin flip in expected goals—4.11 to 4.07—yet they found a way past Vegas anyway. A team that couldn't convert its high-danger chances (0 for 7) scored twice from low-danger ice, the kind of spots most teams leave on the table.

Both sides whiffed on their best looks. Vegas generated 5 high-danger chances and buried none; Buffalo generated 7 and also came up empty in that category. The difference: Buffalo scored twice from lower-danger areas, the kind of spots where most goalies make the save. A. Schmid didn't.

A. Lyon, the Sabres' goalie, was the primary difference. He saved 2.07 goals above what his expected-save percentage forecast, turning away dangerous attempts at a rate that beats the models. A. Schmid played well too (1.11 above expected), but he couldn't overcome the conversion gap.

Vegas crashed harder, generating 9 rebound opportunities to Buffalo's 3, but converted only 1 rebound to Buffalo's 0. The Sabres were more efficient when it mattered. The true disconnect: Vegas created more chaos in front, yet Buffalo controlled the high-danger areas (7 to 5) and won decisively in goaltending performance. Neither team could finish its best chances, so the game hinged on margins elsewhere—rebound discipline and one goalie playing above his expected line.

The model projects 2.5 goals for Vegas, 2.5 for Buffalo, with the Golden Knights at 51% win probability. This is a true coin flip.

Buffalo has momentum: 7-2-1 in their last 10, accumulating points on volume. Vegas is struggling: minus-8 goal differential over the same stretch. That should signal a routine Buffalo victory. The underlying numbers say otherwise.

The Golden Knights dominate possession at even strength: 54.0% expected goals and 52.0% Corsi—they control territorial play, create more chances, and dictate the game's geography. Buffalo's corresponding marks are 50.0% and 49.0%. Vegas should generate more by volume; Buffalo has simply finished those chances at a higher clip.

This gap reveals the core tension. Buffalo's shooting luck masks a disadvantage. Their PDO sits at 1.020, hot enough to juice recent results. Vegas's 0.983 is cold—unsustainably so, given the underlying play. Expect some reversion: Buffalo tending downward, Vegas upward, which narrows the gap between the recent scoreboard and what the numbers predict.

Brandon Saad (0.91 xG/60, 3 goals, 6 assists) is marked inactive, removing a key weapon for Vegas. Tomas Hertl takes on added responsibility: 0.88 xG/60, 5 goals, 17 assists, and 4.3% shooting.

Buffalo counters with Jason Zucker (1.07 xG/60, 14 goals, 15.1% shooting) and Josh Doan (1.01 xG/60, 15 goals, 11.5% shooting). Both convert at elite rates.

Vegas's defensive liability is Rasmus Andersson. Buffalo's is Jack Quinn. In a game this tight, those weak points determine whether this unfolds as a possession-driven Vegas win or a Buffalo upset fueled by shooting variance.

The projection makes sense: two teams pulling in opposite directions. Recent form favors Buffalo; underlying play favors Vegas. That's why the model landed on a coin flip rather than declaring a winner.