Buffalo won 3-2 despite controlling just 1.76 expected goals to Vancouver's 3.82. The Canucks' 68% xG advantage evaporated against A. Lyon's netminding.

Lyon stopped 30 of 32 shots (0.938 SV%) and overperformed his xG by 1.82 goals. T. Demko allowed 3 of 15 (0.800 SV%) and underperformed by 1.24—a cavernous gap for a goalie facing only 15 shots. The Canucks wasted their dominance: 32 shot attempts to Buffalo's 15, three high-danger chances to two. They underperformed their xG by 1.82 goals. Buffalo, meanwhile, scored two from low-danger spots—the kind that exposes goaltending weakness.

The Canucks created three rebound opportunities but converted zero, hemming Buffalo in without finishing. The Sabres were surgical in transition: three goals on twenty rush chances, a conversion rate that punished Vancouver's aggressive up-ice press. Zach Benson's snap shot from the high slot at 14:46 of the second period proved decisive.

R. Dahlin led Buffalo with 1G and 1A in 24:51. T. Thompson added 1G and 1A in 19:06. Q. Hughes registered 1A in 30:03 for Vancouver, the only Canuck to consistently generate danger against Lyon.

This is what 68% xG performance looks like when goaltending variance flips the script: a heist.

The model projects a 2.7-2.5 Buffalo edge with the Sabres at 53% win probability — as close to a pick'em as you'll find.

Buffalo owns the underlying play. At 5v5, the Sabres' 50.0% expected goals share tops the Canucks' 44.0%, and their 49.0% Corsi% reinforces the edge. This is structural control.

Yet Vancouver's 0.974 PDO signals regression incoming — they're running cold and due for positive correction. Buffalo's 1.020 PDO cuts the other way, which means some of their results are unsustainable. When PDO normalizes for both teams, the gap narrows.

The Sabres' edge leans on depth. Jason Zucker leads at 1.07 expected goals per 60 minutes, with 14 goals and 14 assists across 6 high-danger opportunities and a 15.1% shooting rate. Josh Doan sits at 1.01 xG/60 with 15 goals on an 11.5% rate and 18 assists but only 2 high-danger goals — his scoring outruns the underlying play.

Vancouver is thinner. Filip Chytil carries 1.03 xG/60 but has only 3 goals and no assists on 1 high-danger chance with a 15.0% rate. Drew O'Connor, at 0.85 xG/60 with 13 goals and 11 assists, has converted 3 high-danger opportunities at 11.8% — more sustainable but lower volume. Neither generates offense at Buffalo's pace.

The model favors Buffalo. Everything else suggests a coin flip.