The Sabres won a game the underlying metrics almost handed to Detroit. Buffalo overperformed their expected goals by 1.80 — an elite margin of luck or finishing — while Detroit's goaltending gave back what should have been an advantage.

Expected goals went 3.54–3.20 in Detroit's favor. Yet the Sabres scored five goals to four. C. Ellis stopped 24 of 28 shots (0.857 SV%), reaching near-expected performance. J. Gibson gave up 22 of 27 (0.815 SV%), finishing 1.80 goals worse than predicted — a decisive underperformance for a goalie in a one-goal game.

Detroit's case should have been stronger. They held 51% of shot attempts and won the high-danger battle nowhere — chances split 4–4. Yet they scored three goals on twenty-five rush chances, a potent conversion rate. Buffalo answered with four goals on twenty-nine chances, higher volume and higher efficiency. The Red Wings' shooting (50% on close looks) showed poise; the Sabres' rush scoring showed desperation that worked.

Mattias Samuelsson's snap shot from the high slot at 01:05 of overtime broke the tie, finishing a game the Red Wings should have found easier to control. The scoreline masked how much goaltending and rush execution, rather than five-on-five play, decided it.

The model has this pegged as a 2.5–2.6 goal game with Detroit at 51% win probability. You cannot ask for a closer matchup.

Both teams arrive wounded. Buffalo has collapsed to -13 goal differential over their last ten games; Detroit sits at -6. That's not the resume of teams playing well, but the Sabres' fall is sharper.

The underlying numbers reinforce the tossup. Buffalo owns a 50.0% expected-goals share at 5v5 against Detroit's 49.0%; their Corsi splits are identical at 49.0%. Buffalo's sole technical edge is a 1.020 PDO, a sign they've been fortunate—expect some regression. Detroit's 0.985 PDO suggests the reverse: they've run slightly cold and may be due.

Goaltending enters as a variable. Both J. Gibson and U. Luukkonen are on back-to-backs, so watch whether either team reaches for a backup or leans on a tired starter through the second half.

Buffalo leans on Jason Zucker and Josh Doan. Zucker generates 1.07 expected goals per 60 minutes, has scored 14 goals with 14 assists, and owns 6 high-danger tallies—his 15.1% shooting rate is inflated and won't hold. Doan sits at 1.01 xG/60 with 15 goals and 18 assists, though only 2 on high-danger chances; his 11.5% conversion is similarly elevated.

Detroit's top threats are Alex DeBrincat and Marco Kasper. DeBrincat generates 0.88 xG/60, owns 17 goals and 28 assists, but landed only 2 on high-danger looks. His 9.0% shooting rate is reasonable. Kasper trails at 0.81 xG/60 with 7 goals, 8 assists, and 1 high-danger goal—a 6.1% conversion rate that signals good fortune.

The margins here are gossamer. Both teams are struggling, both are generating nearly identical chances, and both arrive with goaltending questions that could determine the outcome either way. Call it what the model sees: a coin flip.