New Jersey won 4-3 in overtime, but the underlying metrics said it should have been closer: expected goals split 2.65–2.22, and high-danger chances finished 4–4. The Devils didn't dominate. They converted.

New Jersey finished 3 of 4 high-danger chances (75%). Montreal converted zero. That disparity didn't come from volume—both teams created four—but from execution in the moment. Three of New Jersey's goals came off rush chances, where the Devils scored 3 on 28 attempts. Montreal, by contrast, scored 2 goals on 17 rush chances while also creating five rebound opportunities off aggressive net crashing, converting one.

Goaltending reflected the gap. J. Dobes gave up four on 28 shots, underperforming his expected goals model by 1.35. J. Markstrom, tested less (three on 19), fell 0.78 short of expectation. Both goalies met quality looks they couldn't keep out. Yet Montreal carved out two goals from low-danger spots—a tax any opposing goaltender pays—while New Jersey's high-danger clinic offset the close underlying metrics.

Jesper Bratt's backhander at 01:33 of overtime settled it. That goal capped a Devils performance that was efficient rather than dominant. The shot chart tilted New Jersey (28–19 on goal), and their expected-goals edge confirmed it. But a 4-3 final demanded execution where Montreal faltered. The real story wasn't a blowout; it was a team that finished when it mattered most against one that didn't.

The model projects a 2.5-2.7 goal spread with the Devils at 54% win probability. This is as close to a pick'em as you'll find.

The underlying numbers confirm it. Montreal holds 48.0% expected goals share and 49.0% Corsi%, compared to New Jersey's 50.0% xG% and 51.0% Corsi%. The gap between these teams barely registers.

The Devils' advantage comes from regression and rest. New Jersey's 0.971 PDO signals sustained bad luck; they're due for positive reversal. Montreal's 1.018 PDO has them running hot, which means regression downward. Add the Devils' three-day rest cushion against Montreal's one day, and the model's slight New Jersey lean becomes coherent.

S. Montembeault enters on the back-to-back, a variable to monitor.

Offensively, the players separate clearly. Cole Caufield (33G, 0.97 xG/60, 20.0% shooting) is Montreal's volume creator and elite finisher. Oliver Kapanen (19G, 0.83 xG/60, 17.8% shooting) operates at lower volume but comparable efficiency. New Jersey counters with Timo Meier (1.08 xG/60, 13G, 3 HD goals, 6.7% shooting), whose expected output substantially exceeds his goal total—another regression candidate. Jack Hughes (18G, 0.88 xG/60, 11.5% shooting) provides secondary scoring.

The tension: can Montreal steal a game on short rest against a Devils team whose luck is reversing? At 54%, the model treats randomness as decisive. That's the definition of a coin flip.