2-1: Final (OTT). 2.27-2.34: xG (OTT-MTL). 10-4: HD chances (OTT-MTL). 2-1 MTL: Walter Cup Final Ottawa came home down 2-0 in the Walter Cup Final and needing a win to keep the series real. For 40 minutes it got a goaltending duel instead. Gwyneth Philips and Ann-Renée Desbiens were both perfect through two periods, and the game had no goal until 7:32 of the third. Then it had three in under twelve minutes: Hayley Scamurra scored first for Montréal, Peyton Hemp tied it for Ottawa, and Rebecca Leslie won it with 56 seconds left. The underlying numbers were close, 2.27 expected goals (xG) , the chance-weighted value of a team's shots, for Ottawa to 2.34 for Montréal. Ottawa was the better team at even strength and Philips was the better goaltender, and that was enough. The Charge win 2-1 and trail the best-of-five 2-1 . A goaltending duel for forty minutes Nothing about the first two periods pointed to a 2-1 finish. The teams traded a quiet, even game and went to the second intermission scoreless. Ottawa had the small edge in the run of play across the night: it won the shot-attempt share (Corsi) 57 percent and the high-danger count 10 to 4 , even though shots on goal finished even at 28 each . Montréal's most dangerous early looks came shorthanded. With Ottawa on a first-period power play, Shiann Darkangelo broke loose for a shorthanded chance the model graded at 0.31 xG , the single biggest look of the game. It did not go in. Laura Stacey had another shorthanded chance worth 0.20 xG a few minutes later. Ottawa's own power play, 0-for-3 for just 0.21 xG all night, never made Montréal pay. 10-9: P1 shots. 8-6: P2 shots. 10-13: P3 shots The third period decided it Hayley Scamurra broke the deadlock at P3 7:32 , finishing a Maureen Murphy feed at even strength. Peyton Hemp pulled Ottawa level at P3 14:30 , with Gabbie Hughes and Ronja Savolainen credited the assists. Then Rebecca Leslie won it at P3 19:04 , off a Sarah Wozniewicz setup, with 56 seconds left in regulation. The shot quality on the goals tells a story worth pausing on. The first two were not good chances. Scamurra's was a 0.05 xG shot and Hemp's a 0.07 : low-percentage looks of the kind that beat a goalie maybe one time in fifteen, and on this night both did. Leslie's winner was the exception. The model priced it at 0.25 xG , the second-biggest chance of the game behind only Darkangelo's shorthanded look, and it was the one goal the shot quality actually predicted. A scoreless duel was decided by two long shots and one real one. Win probability: Ottawa fell to 10 percent The live win-probability model had Ottawa down to about 10 percent while it trailed 1-0 in the third period. That is the trough on the trace below. Hemp's goal pulled it back to a coin flip, and a tied game between two close teams sat near even until the final minute. Leslie's goal sent it to roughly 90 percent, not 100: Montréal pulled Ann-Renée Desbiens for the extra attacker and still had 56 seconds to find an equalizer. It did not. Goaltending: Philips outdueled Desbiens Gwyneth Philips stopped 27 of 28 and finished at +1.34 GSAx . GSAx, goals saved above expected, compares the goals a keeper allowed to the summed xG of the shots she faced. Philips saw 2.34 xG of work and allowed one goal, so she saved well over a full goal more than an average night would have produced. In a game decided by a single goal, that is the margin. Ann-Renée Desbiens was not the problem at the other end. She stopped 26 of 28 for +0.27 GSAx , a touch above expectation, and the two she allowed were a pair of low-xG shots no model would fault her for. This was the reverse of Game 2, where the skaters carried Montréal and the goaltending was a wash. In Game 3 the skaters were close to even and the goaltending was not. The scoring timeline P3 7:32 · Hayley Scamurra · xG 0.05 · OTT 0, MTL 1. P3 14:30 · Peyton Hemp · xG 0.07 · OTT 1, MTL 1. P3 19:04 · Rebecca Leslie · xG 0.25 · OTT 2, MTL 1 Top scoring chances OTT P3 19:04 · Rebecca Leslie default · xG 0.25 (goal). OTT P3 0:40 · Michela Cava default · xG 0.21. OTT P2 8:44 · Gabbie Hughes default · xG 0.17. OTT P2 8:43 · Peyton Hemp default · xG 0.17. OTT P2 4:25 · Rebecca Leslie default · xG 0.16. OTT P3 19:04 · Sarah Wozniewicz default · xG 0.13. MTL P1 14:39 · Shiann Darkangelo default · xG 0.31. MTL P1 16:45 · Laura Stacey default · xG 0.20. MTL P1 7:14 · Catherine Dubois default · xG 0.18. MTL P1 17:46 · Kati Tabin default · xG 0.16. MTL P1 14:39 · Shiann Darkangelo default · xG 0.15. MTL P2 8:22 · Marie-Philip Poulin default · xG 0.12 What this read isn't Not a dominant win. Ottawa took the game, but the xG was a near tie, 2.27 to 2.34, with Montréal nominally ahead. Ottawa's case is an even-strength case: a 57 percent share of the attempts and a 10-to-4 high-danger edge. That is a real edge, but it is a narrow one.. Not a special-teams swing. Neither team scored on the power play (Ottawa 0-for-3, Montréal 0-for-3). It is worth noting where Montréal's danger came from: 0.90 of its 2.34 xG was generated on those three power plays. Ottawa's penalty kill held, and that is part of why the even game stayed even.. Not a series turn. Montréal still leads the best-of-five 2-1. It needs one win in the next two games; Ottawa needs both. A single win narrows the gap, it does not flip the series, and the model still favours Montréal to lift the Cup.. Not a large sample. This is one regulation game, 28 shots a side. Expected goals on that few attempts carries wide error bars, and Philips's +1.34 GSAx is one strong night, not a trend. Game 4 starts from even. 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.. Win probability : the in-game model behind the trace updates after every event from score, time remaining, and game state. The shot map and cumulative-xG charts read straight from the play-by-play.. Numbers regenerate on every build. A PBP re-sync or xG recalibration reshapes every figure and every chart 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/320 . The series projection, including how this Game 3 result moves the Walter Cup Final probabilities, lives at /playoffs/bracket .