The expected goals said close to even—Dallas 3.62, Washington 3.22—but the Stars won. Goaltending decided it.

Oettinger was perfect: 24 saves on 24 shots, a +3.22 versus expected. Thompson was nearly perfect: 18 saves on 19, a +2.62. The difference was that single goal Tyler Seguin scored at 00:43 of the second period from a sharp angle.

Both teams wasted their chances. Dallas generated the sharper looks—0.077 expected goals per shot versus Washington's 0.053—but scored just once on six high-danger chances. Washington had five high-danger chances and scored zero. By the metrics that mattered (expected goals, shot quality, rebounding opportunities), this game was even. The goalies made it lopsided.

Dallas held 44% of shot attempts (19 on goal versus 24) yet stayed competitive in expected goals. That's the asymmetry: fewer shot attempts, better shot quality. Washington threw more pucks at Oettinger but produced lower-percentage looks, then couldn't finish the few dangerous ones they had.

Thompson was particularly stingy on high-danger chances, stopping all six. Oettinger simply didn't need to. He faced clean looks and turned them into nothing. One goal was all it took in a game where expected scoring suggested each team should have put away two or three.

The models can't pick a winner. A 2.5–2.6 goal projection with the Stars at 53% win probability isn't a forecast; it's a shrug. Washington and Dallas have arrived at nearly identical competitive states through different paths, and tonight one team's small edge will either hold or evaporate.

The efficiency numbers reveal why the pick is stuck. Washington leads in Corsi% at 49.0% to Dallas's 47.0%, meaning the Capitals are outworking their opponent at five-on-five—more shot attempts, more volume. But Dallas counters with a 51.0% xG% versus Washington's 50.0%, which means the Stars are converting their chances into higher-quality looks. One team controls possession; the other converts it more ruthlessly. Neither edge is large enough to call dominant.

The complication is sustainability. Dallas is sitting at a 1.024 PDO, suggesting their shooting efficiency is running ahead of expectation; regression is baked in. Washington's 1.018 PDO isn't as stark, but it's still warm. Both teams are due for cold spells, though neither is built to survive one tonight.

The player matchup becomes the tiebreaker. Alex Ovechkin remains Washington's frame—19 goals and 17 assists on a 0.97 xG/60 rate, with two high-danger chances and a 13.8% shooting percentage that suggests he's still converting at a clip above average. Anthony Beauvillier (14G, 10.4% shooting) is Washington's leverage; he's generating 0.89 xG/60, meaning his finishing has been a genuine difference-maker.

Dallas counters with Jason Robertson, who leads in raw production: 25 goals, 21 assists, two high-danger chances, and 12.4% shooting. He's operating at 0.96 xG/60, nearly identical to Ovechkin's volume but in a different system. Michael Bunting has been quieter (8 goals, 0.83 xG/60), but his 7.6% shooting rate hints at untapped potential.

The defensive weak links cut both ways. Declan Chisholm has been Washington's liability; Tyler Myers for Dallas. Neither defense is rated to shut anyone down, so whoever's secondary scoring activates first—whoever cashes a Beauvillier or Bunting chance instead of just generating it—shifts the balance.

This is as close to pick'em as analytics gets. The models aren't hedging; they're genuinely uncertain. That uncertainty is earned.