Both Conference Finals look the same on paper: a road team has stolen home ice and grabbed a series lead. The Habs took Game 1 in Carolina. The Knights swept the opening pair in Denver. If you only read the standings, you'd assume both underdog stories were running on the same fuel.
They are not. Vegas is genuinely outplaying Colorado. Montreal is not outplaying Carolina — they outscored them once, and the gap between those two sentences is the whole story.
West: Vegas Is Really This Good
Series: Golden Knights lead 2-0 — both wins in Denver, 4-2 and 3-1
It would be easy to look at the shot totals and assume Colorado is getting hosed by variance. Colorado outshot Vegas 103–72 across the two games. Surely that's a 2-0 series being held together by hot goaltending? It isn't. Look at where those shots came from:
Nine to one in high-danger chances. The Avalanche didn't generate a single high-danger look through the first two periods of Game 1. They piled up perimeter volume, Vegas absorbed it cleanly, and then in the third — when Colorado finally pushed — Vegas posted 2.71 xG and 5 high-danger chances in a single period on the counter.
The pattern repeated in Game 2. Colorado dominated the second period structurally — 25 shots to 11 — and generated 1.21 xG to Vegas's 0.11. Then Vegas walked into the third and put up 1.81 xG on four high-danger looks. Goodnight.
The goalie split is real but secondary. Carter Hart has been roughly +2.5 GSAx through two games; Scott Wedgewood is around -1.4. Even if you neutralize the goaltending, Vegas was 9-1 at the danger area. That's a structural problem, not a finishing slump.
Bottom line: Colorado has somewhat unlucky finishing (xG suggests 1-2 more goals across the two games), but they are losing the battle that determines series. The model has Game 3 at 47-53 in Colorado's favor, which feels generous given how Vegas has dictated the inside.
East: Carolina Is Better. The Scoreboard Just Hasn't Noticed.
Series: Canadiens lead 1-0 — won Game 1 in Carolina, 6-2. Game 2 tonight.
This is not the Vegas story dressed up in different colors. This is its inverse.
The expected goals were essentially even — Montreal 3.45, Carolina 3.31. Yet the goal differential was four. Where did it come from?
Montreal scored 6 goals on 3.45 xG. That's +2.55 goals above expectation.
Carolina scored 2 goals on 3.31 xG. That's −1.31 below expectation.
Net luck/finishing swing: ~3.9 goals.
The Habs scored four goals in the first period on 1.89 xG. They had seven high-danger chances in the first 20 minutes — that part is real, they can absolutely get to the inside — but the conversion rate was historic.
Then Carolina punched back. In the second period, the Hurricanes put up 36 shots and 1.61 xG against Montreal's 0.40. It was a structural beatdown. They scored once. They should have scored three.
The goalie split:
Jakub Dobeš (MTL): faced 3.31 xG, allowed 2 — roughly +1.3 GSAx. Solid.
Frederik Andersen (CAR): faced 3.45 xG, allowed 6 — roughly −2.5 GSAx. Detonation. Almost all of it concentrated in the first period.
Bottom line: Carolina is the better team. Montreal got the result, not the play. Andersen will not allow 6 on 3.45 xG again barring real decline or injury. The model has Carolina at 81% tonight — that's probably a touch hot for a single game (market has them around 65% no-vig), but the directional read is correct.
What I'm Watching Tonight and Tomorrow
CAR–MTL Game 2 (tonight, Carolina): The single highest-leverage question is whether Montreal can replicate the high-danger volume from Game 1's first period. If they're getting 5+ high-danger chances again, this is a real series and Carolina has a defensive problem. If they're at 2–3 HD chances and Andersen looks normal, it's a 6-2 mirage and Carolina equalizes by Thursday.
VGK–COL Game 3 (Sunday, Vegas): I want to see whether Colorado finally gets meaningful interior chances. If they're back at 0–1 high-danger looks through 40 minutes, this is over in five. If they can crack 5+, the underlying play is shifting and the model's near-coin-flip read is correct.