The Penguins escaped with a 4-2 win despite forfeiting the underlying game. Pittsburgh's expected goals landed at 2.94 to Buffalo's 4.36—a split that reads 40-60 and should have favored the Sabres decisively. T. Jarry stole this one, stopping 29 of 31 shots at a 0.935 save percentage and posting a +2.36 above expected. U. Luukkonen was less fortunate, stopping 15 of 18.

The chasm opens at close range. Buffalo generated eight high-danger chances. They scored zero. Jarry was 8-for-8 on those chances—a shutout in the scoring area. Pittsburgh converted three of their four goals from low-danger territory, the kind of finishes that expose a goaltender's margins. Off the rush, Pittsburgh scored on three of 21 chances while Buffalo buried two of 27.

Pittsburgh won the game Buffalo should have won. The Penguins held 38 percent of shot attempts and 40 percent of expected goals. Connor Dewar's wrist shot from the slot in the third period (19:02) marked the moment Pittsburgh pulled away, but the Sabres' real collapse happened earlier—in Jarry's glove, and in their failure to finish the quality chances he left.

Buffalo underperformed their expected goals by 2.36, a cavernous miss rate. Pittsburgh overperformed by 1.06, half as much but enough. Goaltending decided this, and Jarry's 100 percent save percentage on high-danger chances left no room for interpretation.

The model projects a 2.5–2.7 game with Pittsburgh at 53% win probability. This is about as close to even as the numbers allow.

The underlying metrics confirm it. Buffalo runs 50.0% expected-goal share and 49.0% Corsi%, Pittsburgh 51.0% xG% and 50.0% Corsi%. Both teams are fighting to a draw in possession and shot quality. Buffalo's PDO of 1.020 versus Pittsburgh's 1.012 tells a story: the Sabres are riding hot, and regression is coming.

Jason Zucker is Buffalo's engine. At 1.07 xG/60 with 15.1% shooting, he's produced 14 goals and 14 assists, 6 from high-danger areas. Josh Doan tracks him at 1.01 xG/60 and 11.5% conversion (15 goals, 18 assists), though only 2 came from high-danger areas. Pittsburgh counters with Rickard Rakell (1.01 xG/60, 13 goals, 13 assists, 3 high-danger). But Filip Hallander, who generates 1.28 xG/60, is sidelined pending evaluation after being recalled.

Both teams have defensive vulnerabilities. Jack Quinn (Buffalo) and Caleb Jones (Pittsburgh) are weak links. Here's the odd part: neither team has scored high-danger or rebound goals at any rate, suggesting both goaltending and structure are holding.

The 53% for Pittsburgh is real but narrow. This is a coin flip.