3-2: Final (SEA). 3.49-2.17: xG (SEA-BOS). 10-6: HD (SEA-BOS). +0.17: Schroeder GSAx This game spent most of its 60 minutes looking like a Boston road win. The Fleet led 1-0 after the first period, gave back the equaliser six seconds before intermission, then took a 2-1 lead in the second on a Susanna Tapani 0.26 xG wrister. Seattle's chance-creation profile pushed them ahead at 5-on-5 (Corsi 51%, in-play xG 3.49 to 2.17, high-danger 10 to 6), but the scoreboard didn't move until the third period. Shorthanded tying goal, then a 5-on-5 winner 87 seconds later Down 2-1 with Seattle on the penalty kill, Mikyla Grant-Mentis scored at P3 14:47 on a 0.09 xG backhand to tie the game. 87 seconds of game-clock later , at P3 16:14, Danielle Serdachny added a 0.09 xG 5-on-5 wrister for the eventual game-winner. Two Seattle goals on combined 0.18 xG inside a minute-and-a-half window flipped a one-goal deficit into a one-goal lead. Boston had the remaining 3:46 to find a tying goal and didn't generate the in-tight look they needed. Boston's two regulation goals Theresa Schafzahl opened the scoring at P1 12:30 on a 0.11 xG wrister at 5-on-5; Susanna Tapani added the lead-restoring goal at P2 11:27 on a 0.26 xG wrister, the highest-quality scoring shot of the game by either team. Boston's two scoring shots came on combined 0.37 xG of expectation, finishing roughly 5x expected on a per-shot basis. The Fleet generated six high-danger looks across the night and finished only the Tapani wrister; the rest were saved by Schroeder. Goaltending, decomposed Corinne Schroeder stopped 25 of 27 live shots worth 2.17 xG, finishing at +0.17 GSAx . The two regulation goals against her were the Schafzahl 0.11 xG wrister and the Tapani 0.26 xG attempt; on Boston's other four high-danger chances she made the saves that kept the game in reach for the third-period comeback. Abbey Levy faced 38 live shots worth 3.49 xG and conceded three. Her GSAx came out at +0.49 , an above-expected single-game line on a heavy workload. Two of the three Seattle goals were under 0.10 xG; the save process was much closer to expected than the score line suggests and her overall line on a 38-shot night reads as above-average. The game in cumulative xG The scoring timeline P1 12:30 · Theresa Schafzahl · xG 0.10 · SEA 0, BOS 1. P1 19:54 · Alex Carpenter · xG 0.07 · SEA 1, BOS 1. P2 11:27 · Susanna Tapani · xG 0.24 · SEA 1, BOS 2. P3 14:47 · Mikyla Grant-Mentis (SH) · xG 0.09 · SEA 2, BOS 2. P3 16:14 · Danielle Serdachny · xG 0.09 · SEA 3, BOS 2 Boston's looks above 0.07 xG The Fleet's chances the model graded above the 7% scoring-probability threshold. Two of them went in (Schafzahl and Tapani); the rest are the chances Schroeder turned away. Multiple shots in the 0.10 to 0.20 xG range, all stopped after the second-period Tapani conversion. P2 11:27 · Susanna Tapani (goal) · xG 0.24. P1 15:36 · Susanna Tapani · xG 0.21. P2 3:15 · Jill Saulnier · xG 0.15. P1 17:08 · Alina Müller · xG 0.14. P1 1:23 · Daniela Pejšová · xG 0.14. P1 19:19 · Jill Saulnier · xG 0.13 What this read isn't Not a claim Boston was robbed. They lost the chance-quality battle by 1.5 xG and the high-danger battle 10 to 6. The result reflected the underlying read on the run of play; the Fleet played from in front for 50 minutes but the Torrent's chance-creation profile was always going to find a goal somewhere.. Not a single-game indictment of Abbey Levy . Three goals on a 38-shot workload at +0.79 GSAx is an above-expected save line. The losing variable wasn't her save process; it was the volume she had to face.. Not a claim Seattle's two-goal third is repeatable. Two goals on combined 0.18 xG inside 87 seconds is finishing variance landing in a tight window; the chance-creation profile is the keeper, not the conversion rate.. Not a model failure. The model's chance-quality split favoured Seattle. They won. The compressed scoring run is the kind of single-game pattern variance allows. 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/73 .