Sharks edge the Stars in overtime

This is as close to a pick'em as you'll find. The model projects DAL 2.5 - 2.4 SJS with the Stars at 50% win probability, meaning neither team has a meaningful edge in expected outcome.

The underlying play slightly favors Dallas. The Stars own 51.0% xG at 5v5, while San Jose sits at 48.0%. That's a measurable edge in shot quality, but not decisive — Corsi% is dead even at 47.0% for both teams, suggesting the Stars are converting chances at a higher rate than their volume would normally dictate. Dallas's PDO of 1.024 confirms it: they're running hot, and some regression is likely. San Jose's 0.998 PDO is exactly neutral, meaning they're getting what their underlying play deserves.

The Stars' advantage is real but fragile. Regression alone could swing this game.

On the Dallas side, Jason Robertson remains the team's engine, posting 0.96 xG/60 with 25G 21A and a shooting percentage of 12.4% across 2 HD goals. Michael Bunting offers secondary production at 0.83 xG/60 and 8G 14A, though his 7.6% shooting rate leaves room for variance. Dallas's offensive structure is a warning sign: 0% HD goals and 0% rebound goals means they're generating nothing but primary chances. Their defense faces a liability in Tyler Myers.

San Jose's offensive picture is constrained by injuries. Pavol Regenda, who led their scoring with 1.10 xG/60, 5G 1A, and 2 HD goals at a 16.1% shooting rate, is likely out. Igor Chernyshov, posting 0.91 xG/60, 6G 8A, and 1 HD goal with a 15.0% shooting rate, becomes the default scoring focal point — but he's also listed as San Jose's defensive weak link, creating a dual vulnerability. The Sharks mirror Dallas's offensive limitation: 0% HD goals, 0% rebound goals, another team struggling to generate secondary scoring avenues.

Both teams are one-dimensional offensively and bleeding depth. Dallas has the better foundation, but its PDO advantage is borrowed time. San Jose, despite losing Regenda, remains dangerous because it has no other choice. The Stars project to outscore them by a tenth of a goal—basically variance.