The Stars edged the Sabres 4-3 in a shootout despite losing the xG battle 3.75–2.69. The Sabres controlled 58% of expected goals at 5v5 — this was a heist.

The numbers behind the result. The Sabres held 46% of shot attempts (24–28 on goal). Expected goals finished 3.75–2.69 (58–42% split), and high-danger chances went 6–6.

Variance watch. The Sabres underperformed their xG by 0.75 goals; the Stars overperformed theirs by 1.31.

Turning point. Wyatt Johnston's wrist shot from the high slot, in the shootout.

Goaltending

• J. Oettinger (DAL): 21/24 saves, 0.875 SV% (+0.75 vs expected)
• C. Ellis (BUF): 25/28 saves, 0.893 SV%

Storylines

• DAL stole this one. BUF owned 58% of the xG and lost.
• DAL generated higher-quality looks than the shot total suggests.
• DAL couldn't convert — 0/6 on high-danger chances. That won't happen often.
• BUF was lethal off the rush — 3 goals on 28 rush chances.
• DAL was lethal off the rush — 2 goals on 20 rush chances.
• DAL created 7 rebound opportunities (1 converted) — crashing the net effectively.
• C. Ellis was a wall on HD chances — 100% SV on 6 high-danger shots.

Final: DAL 4 – BUF 3 (SO)

Data: NHL API, MoneyPuck xG model

The model projects a 2.5-2.5 game with Buffalo at 51% win probability. This is as close to a pick'em as they come.

The underlying metrics explain why. Dallas posts 51.0% expected goals share at 5-on-5 to Buffalo's 50.0%. Corsi favors the Sabres slightly, 49.0% to 47.0%. Both teams are running nearly identical PDO — Dallas at 1.024, Buffalo at 1.020 — meaning both are riding unsustainable luck and due for regression.

Goaltending introduces the first variable: U. Luukkonen faces a back-to-back and may not start. If Buffalo opts for a backup, that tilts the matchup considerably. If Luukkonen plays tired, the Sabres lose a structural advantage.

Dallas's offense runs through Jason Robertson. He carries 25 goals and 21 assists with elite xG production (0.96 per 60) and a 12.4% shooting rate that suggests sustainable conversion. Michael Bunting supports at 0.83 xG/60 with only 8 goals, indicating he's due positive regression. Buffalo answers with Jason Zucker, who leads in expected creation at 1.07 xG/60 and has banked 14 goals with 6 high-danger chances. Josh Doan pairs with him at 1.01 xG/60, carrying 15 goals at an 11.5% shooting rate.

Both rosters share a peculiar weakness: neither team converts high-danger chances, posting 0% high-danger goal rates. Both also show 0% rebound goal rates. Dallas's defensive liability is Tyler Myers. Buffalo's is Jack Quinn. These weak links will determine which team escapes the other's scoring chances.

The verdict is as simple as the model suggests: this is a coin flip. Dallas has the edge in expected goal share. Buffalo has the marginally better play-driving rate and the marginal goaltending advantage if Luukkonen is fresh. Regression could cut either way. Bet the line only if it lands on your side of 51%.