The Islanders won 3-2 in overtime. The Devils generated 72% of expected goals. This was a heist by New York.

The Devils controlled the shot map — 59% of attempts, 35 to 24 on goal — but the puck traffic didn't translate to the scoreboard. Expected goals finished 4.69 to 1.86, a 72-28 split: New Jersey was substantially more dangerous by every process metric. They generated 7 high-danger chances to the Islanders' 1, creating looks at a higher caliber — 0.063 xG per shot versus 0.048. The Devils should have won this game.

They didn't, because of I. Sorokin. The Islanders' goaltender stopped 33 of 35 shots (0.943 SV%), a +2.69 overperformance that carried overtime. He was a wall on high-danger chances — 100% save percentage on the 7 high-danger shots the Devils generated. J. Markstrom faced easier volume — 21 of 24 shots — yet finished at 0.875 SV%, underperforming his xG by 1.14. The Devils had elite process. Their goaltender didn't match it. The Islanders' goalie wouldn't budge.

The Devils could not finish. Zero goals on seven high-danger chances is variance that doesn't repeat. The Islanders capitalized on what they got: 2 goals on 14 rush chances. Mathew Barzal's overtime winner at 01:17 ended a game New Jersey should have closed out.

New Jersey crashed the crease hard — 7 rebound opportunities created, 0 converted. The Islanders created 3 and converted 1. That marginal efficiency, paired with Sorokin's brilliance, determined the outcome.

The model projects a 2.5-2.7 goal result with New Jersey at 54% win probability. This is about as close to a pick'em as you'll find.

New Jersey has controlled play. The Devils generated 50.0% of expected goals and possessed 51.0% of shot attempts—both edges over the Islanders' 48.0% and 49.0%. But the Devils are running cold. A PDO of 0.971, a measure of puck luck and bounces, says variance is against New Jersey. The Islanders' 0.993 means regression favors the Devils. In a 2.5-goal projection, luck decides it.

I. Sorokin plays on back-to-back. Monitor whether the Islanders deploy a backup or risk their starter fatigued—that choice could tilt a game where a single goal is the margin.

The edge goes to whoever shoots cleaner. Anders Lee generates 1.20 xG/60 and has 9 goals on 6.4% efficiency; Jack Hughes matches that intensity with 18 goals at 11.5%. Bo Horvat (18 goals, 12.8%) is the Isles' deadliest scorer. Timo Meier brings 13 goals and a 6.7% rate for the Devils. Lee has 2 high-danger goals, Meier 3—both signs of clinical finishes—yet neither team has scored from high-danger or rebound chances all season. Zero percent from in-close means every goal lives in structure or luck. The team that shoots better from distance and limits defensive breakdowns wins.

Alexander Romanov (Islanders) and Johnathan Kovacevic (Devils) are defensive weak points. When either falters, the opponent's finishers strike.

A genuine coin flip. Take the model's 54% and hold it.