0-1: Final (MTL). 20-19: Shots (BOS-MTL). 1.30-1.01: xG (BOS-MTL). +1.30: Desbiens GSAx This game produced 2.31 combined xG across sixty minutes of regulation hockey. The league-average regulation game runs in the 4 to 5 xG range; this one didn't get halfway there. The shot totals were close (Boston 20, Montreal 19), the territorial possession share barely budged from 50-50 (Boston Corsi 51%), and high-danger chances were single-digit on both sides. The model graded this match-up as a near coin flip on process. Goaltending and one finished mid-quality chance decided the outcome. The only goal Montreal's goal arrived at 5:52 of the third period from Lina Ljungblom at 5-on-5, with primary assist Laura Stacey and secondary assist Nicole Gosling . The model priced Ljungblom's shot at 0.17 xG , a normal mid-quality look, the kind that scores roughly one time in six. There is no in-game outlier here: this was not a broken-coverage two-on-zero or a deflection from a metre out. It was an average shot that beat an average save expectation, and on this particular night that was the entire margin. Worth flagging the immediate context. Boston was on the power play until 5:41 of the third (a Laura Stacey tripping minor that expired without a goal), and the Montreal goal came at 5:52, eleven seconds into even strength. That puts the scoring play at the back end of a Boston power-play formation reset, which is a useful piece of context for the next section. Goaltending, decomposed Ann-Renée Desbiens faced 20 shots worth 1.30 xG and conceded zero. That is +1.30 goals saved above expectation in a single game. For context, the per-game GSAx leaders across a full PWHL season live in the +0.4 to +0.6 range. A +1.4 night is well into outlier territory; in a 1-0 regulation game played on this shot volume, it is the actual margin of victory. Aerin Frankel faced 19 shots worth 1.01 xG and gave up one. Her GSAx for the night was +0.01 , which is to say the model called her workload almost exactly. She was not the reason Boston lost. The model graded both goaltenders favourably; Desbiens was simply two points of save percentage and one full goal of expectation better. The game in cumulative xG The scoring timeline P3 5:52 · Lina Ljungblom · xG 0.15 · BOS 0, MTL 1 Boston's power-play problem The Fleet had six power-play opportunities . They generated 2 shots and 0.13 xG across all of them combined. An average PWHL power play of that length should clear the 0.4 to 0.5 xG mark; getting 0.13 from twelve combined minutes of 5-on-4 time is the kind of unit-level output that explains a 1-0 loss without needing to look anywhere else. Whatever the sequence breakdown was (entries, formation set-up, shot-pass selection), the model is saying the first dangerous look almost never came. Boston's best non-goals The chances the model graded as Boston's best looks at Desbiens. Every entry here is a shot the average PWHL shooter converts at least 7% of the time. The headline miss is the one at the very end of regulation. P3 18:34 · Abby Newhook · xG 0.23. P1 6:34 · Jessie Eldridge · xG 0.10. P1 14:10 · Liz Schepers · xG 0.09. P3 1:37 · Loren Gabel · xG 0.08. P1 1:09 · Liz Schepers · xG 0.08. P1 14:04 · Liz Schepers · xG 0.07 What this read isn't Not a claim Boston was the better team and got robbed. They weren't, and they didn't. xG was within 0.9 of even, both teams produced single-digit high-danger counts, and Ann-Renée Desbiens was the analytical difference. Boston's offence did not earn a goal off in-game play.. Not a claim the power play is broken. Six opportunities is a small sample. One night of 0.14 xG against a structured kill is not a unit-level diagnosis; it is one bad sample worth flagging and moving past.. Not a single-goalie indictment of Aerin Frankel . One goal on 1.06 xG is essentially a flat GSAx night. The goalie-on-goalie comparison is not where the loss lives.. Not a model failure. A 1-0 game with both sides under 1.5 xG and one shot beating its expected probability is well within the normal variance of single-game outcomes. xG predicts process, not which mid-quality shot finds the back of the net on a given Saturday. 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/105 .