Expected goals said 3.88–3.90 — dead even — but Washington's 6–2 final scoreline told a different story. The underlying play was close. The execution wasn't.

Washington's dominance came from two places: goaltending and finishing. L. Thompson stopped 37 of 39 shots (0.949 save percentage), beating his expected mark by 1.90 goals. A. Lyon let in 2 of 5 (0.400 save percentage), falling 2.31 short. That four-goal swing between the pipes was the difference between a one-goal game and a blowout.

The Capitals overperformed their expected goals by 2.12, converting chances at an elite rate. Buffalo did the opposite — underperforming their xG by 1.90, unable to finish when it mattered. The Sabres created higher-quality looks than their shot total (28–39) suggested, converting just 0 of 5 high-danger chances. Washington converted 9 of those, a lopsided 9–5 edge.

Tom Wilson's backhander from the slot (third period, 8:55) was the turning point, pushing Washington clear when the game remained contested.

Washington was lethal on the rush, scoring 4 goals on 29 chances. Buffalo matched that rush-goal total (2 on 30 attempts), but couldn't replicate it enough to keep pace. The Capitals crashed the net effectively, generating 4 rebound opportunities and converting 2; Buffalo generated the same 4 rebounds and converted none.

The model has this pegged as precisely even: Capitals at 52% win probability, the projection running to 2.6–2.7 goals for Buffalo. Neither team possesses a decisive edge in the metrics that matter.

Both squads sit at identical 50.0% expected goals shares and 49.0% Corsi percentages at five-on-five. The separation is negligible. What matters instead is how close both are to regression. Buffalo's PDO stands at 1.020; Washington's at 1.018. Both are running hot enough that some regression is due, and whenever that happens, whoever gets unlucky first loses the cushion.

The forward matchup tilts slightly toward Buffalo. Jason Zucker is driving play at 1.07 xG/60 and has already banked 14 goals with 14 assists and six high-danger chances; he's shooting 15.1%, which is sustainable given his volume. Josh Doan sits just behind him at 1.01 xG/60 and has scored 15 times with 18 assists, though he's riding a softer 11.5% on only two high-danger goals—some of that outperformance may not stick.

Washington counters with Alex Ovechkin, who despite posting 19 goals and 17 assists is actually generating slightly less (0.97 xG/60). That gap between results and creation matters; Ovechkin is at 13.8%, which is elevated. Anthony Beauvillier (14G, 9A) checks in at 0.89 xG/60 and 10.4% shooting, meaning both Capitals scorers are overperforming their underlying volume.

The defensive concern cuts both ways. Buffalo has no high-danger goals and no rebound goals—that's not defensive excellence, it's either noise or genuine weakness in transition recovery. Jack Quinn is the noted defensive liability. Washington mirrors the pattern: zero high-danger goals, zero rebound goals, with Declan Chisholm identified as vulnerable. Both teams are leaning on goaltending to hold together a defensive picture that hasn't been stress-tested.

The edges cancel out. Washington's slight win-probability advantage reflects home ice and the recency of Ovechkin's goal total, but the models know those Capitals scorers are running hot. Buffalo has the steadier underlying advantage in shot creation, yet both teams are overdue for some PDO gravity.

This is as close as games get. The team that avoids the first regression stumble wins.