The Oilers won 4–3 despite the Avalanche controlling the game. Colorado dominated shot attempts (56%) and expected goals (3.52–3.25), but Edmonton's goaltending and finishing made the difference.

The divergence was stark. The Avalanche generated more chances of better quality—0.071 expected goals per shot versus Edmonton's 0.050—yet scored fewer goals. This is a game Colorado should have won on the numbers.

T. Jarry stole this one. The Oilers' primary starter (11 saves, 0.917 SV%) was perfect on high-danger chances: 100% save percentage on six HD shots. By the model's measure, he saved 2.5 goals above expected. That's a wall.

The Avalanche, meanwhile, shot themselves in the foot. They generated six high-danger chances and converted zero. Low-danger opportunities went two ways—Colorado buried two from that range—but those aren't wins the data rewards. M. Blackwood allowed 0.75 goals above expected, getting no help from his defense when it mattered most.

Edmonton's efficiency told the story. Two of five high-danger chances converted—exactly 40% of the looks they got. Colorado created four rebound chances, crashed the net effectively, and got nothing. That's execution under pressure, the kind that decides playoff-caliber games.

There was a turning point: Connor McDavid's snap shot from the slot (P3, 09:03). One moment, one decision, and the game's momentum shifted. McDavid finished with a goal and an assist. E. Bouchard added two assists; L. Draisaitl collected two as well.

The model projects a 2.5-2.9 Avalanche victory with Colorado holding 57% win probability. This lean rests on superior underlying play and looming statistical regression.

Colorado dominates at five-on-five: 57.0% expected goals against Edmonton's 52.0%, a 5-point edge that reflects territorial control. Corsi mirrors it—57.0% for Colorado, 50.0% for Edmonton. The Avalanche will dictate possession.

The complication is PDO regression. Colorado's 1.021 PDO runs hot, banking bounces and shooting efficiency that historically mean-revert. Edmonton's 0.982 PDO runs cold. Expect Colorado's edge to compress as Edmonton's underlying luck improves.

Zach Hyman anchors Edmonton's offense at 1.23 expected goals per 60 minutes, producing 14 goals and 13 assists with 1 high-danger goal. His 13.0% shooting rate indicates genuine finishing prowess. Connor McDavid's 0.92 xG/60 is lower—a volume difference—yet he converted it into 25 goals with a 13.7% shot rate and 5 high-danger tallies. His efficiency remains elite.

Colorado counters with Artturi Lehkonen at 1.11 xG/60. He has 18 goals and 22 assists with 6 high-danger goals at 14.9% shooting, signaling dangerous finishing. Ross Colton operates at 1.10 xG/60 with 7 goals and 15 assists; his 4.9% shooting rate suggests positive regression ahead.

Structurally, Jake Walman is Edmonton's defensive liability, while Nazem Kadri fills that role for Colorado. Colorado's territorial dominance means Edmonton defends more frequently—a structural disadvantage when territory controls play.

The model's thesis: underlying dominance survives PDO normalization. History suggests it usually does.