3-1: Final (MTL). 2.58-2.92: xG (MTL-NY). 9-8: HD (MTL-NY). +1.92: Desbiens GSAx This game spent 47 minutes looking like a New York steal. Casey O'Brien 's 0.27 xG wrister at P1 9:09 was the highest-quality scoring shot of the entire game by 0.10 xG over anything that came after, and it held up as the only goal through the first two periods plus most of the third. New York carried higher xG into the second intermission (2.92 eventual to Montreal's 2.58) and a slightly larger high-danger count. Then Montreal scored three goals in the next 5:35 of game-clock to flip the result entirely. Three goals in 5:35 Catherine Dubois opened the comeback at P3 1:54 with a 0.07 xG snap shot. She made it 2-1 Montreal at P3 6:40 with a 0.10 xG snap shot. Forty-nine seconds later, Laura Stacey tipped a 0.12 xG attempt at P3 7:29 for the eventual game-winner. Combined: 0.31 xG of expectation across the three scoring shots, three actual goals . None of the chances was the kind of in-tight high-danger look the model considers near-coin-flip; Montreal converted on three normal mid-quality attempts in a row while Desbiens kept the New York end clean. Desbiens, decomposed Ann-Renée Desbiens stopped 23 of 24 live shots worth 2.92 xG, finishing at +1.92 GSAx . The O'Brien goal on the only chance the model graded above 0.20 xG was the lone marker against her. Eight Sirens high-danger chances did not beat her. The game's result is the linear convolution of two effects: Desbiens saving more than expected at one end, and Montreal finishing at a normal rate at the other in a compressed window. Kayle Osborne faced 29 live shots worth 2.58 xG and conceded three. Her GSAx came out at -0.42 , essentially the model calling her save process roughly average. None of the three goals against her was over 0.13 xG, and the compressed timing of all three is the kind of single-game pattern that lands on the goalie's ledger but doesn't speak to the underlying save process. New York's chance-quality ceiling O'Brien's 0.27 xG opener wasn't just the New York high; it was the highest-xG scoring shot of either team's night by a 0.10 margin (the next-best scoring shot was 0.13). The Sirens' attack profile was shaped around it: a single in-tight wrister early, then a long tail of mid-quality looks for the rest of regulation, none of which beat Desbiens. They also went 0-for-5 on the power play, with no PP attempt cracking the 0.20 xG threshold. The game in cumulative xG The scoring timeline P1 9:09 · Casey O'Brien · xG 0.25 · MTL 0, NY 1. P3 1:54 · Catherine Dubois · xG 0.08 · MTL 1, NY 1. P3 6:40 · Catherine Dubois · xG 0.10 · MTL 2, NY 1. P3 7:29 · Laura Stacey · xG 0.13 · MTL 3, NY 1 New York's looks above 0.08 xG The Sirens' chances the model graded above the 8% scoring-probability threshold. The O'Brien opener tops the list; the rest are the looks Desbiens turned aside. The shape of the distribution is the analytical evidence for the chance-quality story: NY had real looks all night, just one of them with the in-tight geometry that scores at high rates. P2 16:41 · Casey O'Brien · xG 0.38. P3 4:48 · Paetyn Levis · xG 0.29. P3 4:49 · Paetyn Levis · xG 0.26. P1 9:09 · Casey O'Brien (goal) · xG 0.25. P1 14:18 · Kristin O'Neill · xG 0.18. P3 5:41 · Maja Nylén Persson · xG 0.14 What this read isn't Not a claim Ann-Renée Desbiens is on this trajectory every night. A +2.0 GSAx start is in the top 1-2% of single-game appearances. The right diagnostic is the rolling save-percentage profile across her next several games, not extrapolation from one outing.. Not a claim Kayle Osborne played poorly. A near-zero GSAx night against three goals all under 0.13 xG is a clean save process. The compressed scoring window is mostly variance landing on her side; the underlying performance was much closer to expected than the score line suggests.. Not a claim New York deserved a different result. They generated higher xG than Montreal but the chances were one in-tight look and a long tail. The Charge converted three mid-quality chances in a stretch where the Sirens didn't get a comparable second high-danger opportunity.. Not a claim Montreal's third-period scoring run is a sign of a hot streak. Three goals in 5:35 on combined sub-0.30 xG is single-game finishing variance; it doesn't repeat on demand and shouldn't be folded into expectations for the next game. Methodology xG model : PWHL-calibrated logistic regression on 16,709 non-EN shots. Same fit as the methodology post . Features: distance, angle, rebound (≤3 s), penalty shot, power play, overtime, scorer-tagged quality chance, seven shot-type dummies. Empty-net shots are priced outside the main model (MoneyPuck / NST convention).. GSAx : summed xG on shots faced minus goals allowed, computed per-goalie from the PBP feed. Empty-net shots and goals are excluded from both sides of the ledger.. Numbers regenerate on every build. A PBP re-sync or xG recalibration reshapes every figure in this post on the next static export. Full play-by-play, shift overlays, and the per-player game log for this game are at /games/90 .