Winnipeg won a game it had no business winning. The Sabres controlled 64% of expected goals at even strength, yet lost 4–1. One goaltender's near-perfect game decided it: E. Comrie stopped 3.88 goals above expected.

The Sabres generated 4.88 xG to Winnipeg's 2.74, a 64%-36% split that should produce a narrow result, not a blowout. Buffalo created 9 high-danger chances; Winnipeg created 1 and converted it. The Jets were surgical on the rush, scoring 3 goals on 19 chances. Buffalo crashed the net for 9 rebound opportunities and buried one.

E. Comrie stopped 34 of 35 shots—a 0.971 save percentage—and was perfect on high-danger looks: 9-for-9. U. Luukkonen, by contrast, allowed 19 of 22, posting a 0.864 SV%. The Sabres underperformed their xG by 3.88 goals; Winnipeg overperformed by 1.26.

Gabriel Vilardi opened the score with a wrist shot from the top of the circles, an early goal that tilted momentum. Buffalo's efficient play disintegrated after that—they outshot Winnipeg 35–23, generated more chances at every level, and lost anyway.

The result: a 4–1 scoreline that flattens what was a dominant Buffalo performance into an afterthought. One goalie played near-perfectly. That decided everything.

This is about as close to a pick'em as you'll find. The model projects a 2.5–2.6 goal spread with the Jets at exactly 50% win probability—essentially a coin flip.

Buffalo is the hotter team, racing to 6-4-0 over their last 10. Winnipeg is underwater, posting a -9 goal differential in the same span. Yet the season's 5v5 numbers suggest these teams are functionally identical: Buffalo's xG% sits at 50.0% to Winnipeg's 48.0%, while both teams sit at 49.0% Corsi%. The real divergence is in the margins. Buffalo's PDO of 1.020 is running hot relative to Winnipeg's 0.998—a gap that whispers regression.

Both E. Comrie and C. Ellis are on back-to-backs, so monitor the net for lineup adjustments or tired starters.

The scoring edge belongs to Buffalo's Jason Zucker, who is generating 1.07 xG/60 while carrying 14 goals and 14 assists, with 6 high-danger opportunities. Josh Doan runs a similar offensive profile at 1.01 xG/60, having posted 15 goals and 18 assists on an 11.5% shooting rate. On the other side, Kyle Connor (25 goals, 30 assists) offers higher volume but lower generation at 0.83 xG/60, while Nino Niederreiter (0.87 xG/60, 6 goals, 6 assists) sits squarely in the middle—enough offensive heat to matter, not quite enough to separate.

The Sabres' PDO is the only structural signal pointing anywhere: that 1.020 suggests some of Buffalo's points pace comes from variance, and variance reverts. Everything else—the shot rates, the pace, the personnel—argues for exactly what the model says it is: a toss-up.