Pittsburgh's shutout masks a game where underlying metrics suggested the Penguins deserved better. They generated 3.79 expected goals and seven high-danger chances but finished with zero — a complete variance from their underlying play.

Washington's superiority was genuine. They controlled 57% of five-on-five expected goals (5.05–3.79), led shot attempts 27–24, and created far more dangerous looks — 0.092 xG per shot versus Pittsburgh's 0.058. The Capitals executed on the rush, scoring twice on transition chances.

Still, Washington benefited from fortune. They underperformed their expected goals by 2.05, scoring three on 5.05 xG. L. Thompson allowed none: 24 saves on 24 shots, a +3.79 performance above expected that proved decisive. S. Skinner stopped 24 of 26 for Pittsburgh, posting a +3.05 save percentage above expected, but Pittsburgh never converted when it mattered.

Connor McMichael's snap shot from the doorstep at 17:16 of the third period opened the game. M. Fehérváry recorded two assists. T. van Riemsdyk scored once.

The model sees this as a toss-up: Washington at 51% win probability, a 2.6- to 2.7-goal game. That's not a clear favorite—it's even.

The underlying shot metrics confirm it. Pittsburgh is generating 51.0% of expected goals at 5-on-5, Washington 50.0%; Pittsburgh leads Corsi 50.0% to 49.0%. Both teams exist in a narrow band of competence—neither is running away, neither is collapsing.

The wrinkle: Washington's PDO of 1.018 suggests the Capitals are running hot, converting more than expected. Pittsburgh's 1.012 is closer to neutral. If Washington regresses even slightly, that one-goal separation in the model shrinks further. Watch for a tighter game than the raw xG might suggest.

Goaltending creates a second layer of risk. Both L. Thompson and A. Silovs are on back-to-backs. That doesn't guarantee a backup or a tired starter, but it narrows the margin for error. Fatigue compounds fast in a close match.

On offense, Rickard Rakell is the clearer focal point for Pittsburgh. At 1.01 xG/60 with 13 goals and 13 assists, he's producing volume without elite finishing—his 12.1% shooting rate is elevated but sustainable for a player of his skill. Alex Ovechkin carries Washington's load: 0.97 xG/60, 19 goals and 17 assists, with a 13.8% shooting rate that reflects his elite release. Ovechkin's power play activity is the likeliest source of separation in a close game.

Anthony Beauvillier fills the secondary role for Washington, generating 0.89 xG/60 with 14 goals and 9 assists at a 10.4% shooting rate. He's outperforming his expected goal creation but with less firepower than Rakell.

Filip Hallander (1.28 xG/60) is expected to be unavailable pending further evaluation.

This is what 51% looks like: marginal separation, overlapping quality, a result decided by conversion rate and goaltending performance on a tired night.

Three things to watch:

• Thompson and Silovs both on back-to-backs—fatigue could materialize as goalie weakness.
• Washington's 1.018 PDO—expect regression toward neutral conversion rates.
• Rakell (12.1% shooting) and Ovechkin (13.8% shooting)—margin-of-error finishers in a margin-of-error game.