0-1 OT: Final (TOR-NY). 21-33: Shots (TOR-NY). 1.26-2.17: xG (TOR-NY). +2.43: Combined GSAx Two goaltenders saved a combined +2.43 expected goals above league-average on this workload. On the Toronto side, Raygan Kirk faced 33 shots worth 2.17 xG and allowed one. On the New York side, Kayle Osborne faced 21 shots worth 1.26 xG and allowed zero. Both lines sit in the top few percent of single-game starts by GSAx. The entire two-hour result hinges on the one shot Kirk did not stop: an Allyson Simpson wrister from the point at P4 2:42, priced at 0.16 xG , a look the average PWHL shooter converts roughly one time in six. The chance-creation gap was real New York outshot Toronto 33 to 21 and out-xG'd them 2.17 to 1.26 . At the high-danger level the split was 6 to 2 in favour of the Sirens. Per-shot quality was 0.066 for New York and 0.060 for Toronto, so New York not only took more shots but took slightly better ones on average. Five-on-five Corsi came out 53.7% NY against 46.3% TOR . On every volume-or-quality measure the model produces, the Sirens were the side that should have scored first. They did not. Through sixty minutes of regulation the scoreboard read 0-0 despite New York having already generated 1.97 xG on 29 shots (their regulation share of the 2.17 total). On an average night that workload cashes roughly twice. Kirk turned all of it aside, including a Paetyn Levis 0.13 xG short-handed chance at P3 5:44 and a Sarah Fillier 0.12 xG short-handed look thirteen seconds earlier. New York's two best chances of the entire night came in the first fifteen seconds of overtime. Osborne's shutout, decomposed Kayle Osborne saw 21 shots worth 1.26 xG . On average a PWHL goaltender handed that workload concedes about 1.3 goals. She conceded none. Her +1.26 GSAx is roughly the tail that a season-leading per-game GSAx figure lives in, compressed into a single start. Toronto's best look came on the opening exchange of the third: Savannah Harmon at P3 0:59 priced at 0.14 xG at 5-on-5. The second-best was a Hanna Baskin OT attempt at P4 2:24, 0.13 xG, eighteen seconds before the Simpson winner went in the other direction. Kirk's +1.21 was almost enough Raygan Kirk came into this game off a +1.92 GSAx shutout against Minnesota two nights earlier, a line reviewed on /notebook/game-112 . She posted another multi-xG night here: +1.17 GSAx on 33 shots against. Kirk faced five New York attempts priced at 0.10 xG or higher and stopped all five. The one that beat her was number six: Allyson Simpson from the point at P4 2:42, a 0.16 xG wrister at 5-on-5. Back-to-back +1.2 GSAx starts against top-of-table opponents sits well outside the sustainable per-game range; the rolling reference point for a league-leading PWHL goaltender is closer to +0.5 per start. The game in cumulative xG The scoring timeline P4 2:42 · Allyson Simpson · xG 0.16 · TOR 0, NY 1 Overtime in five shots The five overtime shots captured most of the game's highest-quality chances in a 2:42 window. New York opened with a Jaime Bourbonnais 0.24 xG look at P4 0:11 (stopped) and a Anne Cherkowski 0.14 xG attempt three seconds later (stopped). Toronto answered with a Hanna Baskin 0.13 xG chance at P4 2:24 (Osborne stopped). Then Allyson Simpson from the point, 0.16 xG, beat Kirk. OT xG came out to 0.57 NY to 0.12 TOR across the five shots, compressing the regulation pattern into under three minutes. P4 0:11 · NY · Anne Cherkowski · xG 0.13. P4 0:14 · NY · Jaime Bourbonnais · xG 0.24. P4 2:07 · NY · Kristin O'Neill · xG 0.05. P4 2:24 · TOR · Hanna Baskin · xG 0.12. P4 2:42 · NY · Allyson Simpson (goal) · xG 0.16 Special teams decided nothing Toronto took four minors to New York's two. The Sirens drew 4 power-play opportunities and generated 8 shots worth 0.58 xG , most of it dispersed across the perimeter. They scored none of it. Toronto's 2 opportunities produced 3 shots for 0.24 xG , also without a goal. Combined PP conversion was 0 for 6 . The game was decided at five-on-five in overtime, on a 0.16 xG shot, in spite of two power-play windows the Sirens could have closed the game on earlier. The penalty timeline P1 12:04 · NY · Maddi Wheeler · Holding (2 min). P1 12:48 · TOR · Ella Shelton · Roughing (2 min). P1 16:54 · TOR · Ella Shelton · Illegal Body Checking (2 min). P2 7:03 · TOR · Claire Dalton · Hooking (2 min). P3 4:33 · TOR · Natalie Spooner · Slashing (2 min). P3 11:11 · NY · Maddi Wheeler · Tripping (2 min) New York's looks above 0.08 xG Every Sirens attempt the model graded at 0.08 xG or higher. The top line is the Bourbonnais OT chance (0.24); the Simpson goal clears the list at 0.16. Kirk turned aside every other entry, including two short-handed looks within thirteen seconds of each other in the third period. P4 0:14 · Jaime Bourbonnais · xG 0.24. P4 2:42 · Allyson Simpson (goal) · xG 0.16. P3 2:57 · Maddi Wheeler · xG 0.15. P4 0:11 · Anne Cherkowski · xG 0.13. P3 5:44 · Paetyn Levis · xG 0.12. P3 5:31 · Sarah Fillier · xG 0.12. P2 0:28 · Sarah Fillier · xG 0.10. P3 0:45 · Casey O'Brien · xG 0.09. P1 17:51 · Paetyn Levis · xG 0.09. P3 6:35 · Jaime Bourbonnais · xG 0.09 Toronto's best non-goals P3 0:59 · Savannah Harmon · xG 0.14. P4 2:24 · Hanna Baskin · xG 0.12. P3 18:25 · Daryl Watts · xG 0.09. P1 12:47 · Ella Shelton · xG 0.09. P3 12:21 · Renata Fast · xG 0.08. P3 1:58 · Clara Van Wieren · xG 0.08. P2 14:21 · Kiara Zanon · xG 0.07. P2 19:51 · Sara Hjalmarsson · xG 0.07 What this read isn't Not a claim Toronto deserved the win on chance creation. They did not. The xG gap was roughly a full goal in New York's favour and the Corsi and high-danger splits both lean Sirens. Kirk's GSAx was close enough to carry the game to OT but the chance-creation evidence is unambiguous.. Not a repeatable Osborne forecast. A +1.27 GSAx shutout is a tail start, not a baseline. The same workload on a different night concedes about 1.3 goals. Single-game GSAx has wide variance; the right test for goaltender talent is the season-long save-percentage profile against expected.. Not an indictment of New York's finishing. Going 1-for-33 on a 2.21 xG night is poor conversion, but conversion rates mean-revert hard. The Sirens generated the chance profile you want. Finishing is the variance layer on top, and they drew the bad side of it for fifty-nine minutes.. Not a claim the model underrated short-handed chances. The two short-handed looks in P3 were priced at 0.12 and 0.13, which is consistent with how the model treats 4-on-5 shots. SH shots do not get a systematic bump; they are evaluated on location, angle, and rebound sequence like any other. 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 or less), penalty shot, power play, overtime, scorer-tagged quality chance, seven shot-type dummies. Empty-net shots are priced outside the main model.. GSAx : summed xG on shots faced minus goals allowed, computed per-goalie from the PBP feed. Empty-net shots and goals are excluded on both sides.. Corsi and high-danger splits : computed at five-on-five only. Power-play shots land in the PP xG columns, not the Corsi or high-danger columns.. 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/113 .