Pittsburgh won 7-2 despite ceding 63% of expected goals. A. Silovs turned the game into a robbery.

Colorado controlled play at 5v5, holding 57% of shot attempts and 63% of xG (3.92–2.26). The Avalanche generated six high-danger chances to Pittsburgh's four. By every reasonable measure, they played better. They lost.

This was a goaltending mismatch in slow motion. Silovs stopped 1.92 goals above his expected save percentage, making 25 of 27 shots look routine. M. Blackwood, Colorado's primary starter, surrendered 2.30 goals more than expected and allowed 11 of 15 in limited action before S. Wedgewood took over and lasted exactly two saves on five shots (0.400 SV%).

Pittsburgh converted at an elite clip. E. Malkin scored twice on 16:18 of ice time. E. Karlsson added a goal and two assists in 21:44, controlling play from the perimeter. B. Rust chipped in a goal and assist. The three combined for five of Pittsburgh's seven goals—exactly the efficiency Colorado lacked. The Avalanche managed 0-for-6 on high-danger chances while Pittsburgh went 2-for-4 in those spots. That margin doesn't hold in rematches.

Noel Acciari's wrist shot from the doorstep in the third period was the turning point, arriving when momentum had already tilted sharply. Pittsburgh's rush game was lethal: five goals on 19 rush chances showed how thoroughly they punished Colorado's attempts to push pace.

Colorado underperformed their expected goals by 1.92—a sign that even by their underlying metrics, they couldn't finish. Pittsburgh overperformed by 4.74. Elite goaltending and ruthless execution in the other direction turned a closer game into a rout.

The model projects a 2.5–2.9 goal Avalanche win, pricing Colorado at 57% to take it.

Colorado's edge starts with control. The Avalanche dominate 5v5 expected goals at 57.0% to Pittsburgh's 51.0% — a meaningful margin that translates to territorial advantage. Corsi reinforces it: 57.0% to 50.0%, meaning Colorado generates more shot attempts and dictates play.

But regression lurks. Colorado's PDO sits at 1.021 — the team is running hot — while Pittsburgh's 1.012 suggests some cooling could come. That PDO gap could compress.

Pittsburgh will lean on Rickard Rakell. He's scored 13 goals on 12.1% shooting, with three from high-danger spots. But Filip Hallander is likely unavailable — he was recalled from the WBSPenguins (AHL) for further evaluation. He had generated 1.28 xG/60 on just 0 goals and 3 assists before his status clouded.

Colorado has depth. Artturi Lehkonen has scored 18 goals on 14.9% shooting, with six from high-danger areas on 1.11 xG/60 — he's converting at a premium to his underlying shot quality. Ross Colton, at 1.10 xG/60, has 7 goals but only 4.9% shooting, leaving room for improvement.

Defensively, both teams show cracks. Pittsburgh's Caleb Jones is a clear weak link; Colorado's Nazem Kadri mirrors that vulnerability. Neither team converts high-danger chances — both sit at 0%. Neither has solved that problem.

The model's 57% Colorado lean reflects superior process: better possession, better shot volume, better expected-goal creation. PDO regression is real, but insufficient to overcome Colorado's underlying edge.