Goal-scoring in end-of-game NHL hockey doesn’t follow a Poisson process. We apply a Weibull competing-risks hazard model to recent play-by-play data to study event timing — and optimal pull strategy — in goalie-pulled windows.

• Scraped recent NHL play-by-play data to isolate goalie-pulled windows and distinguish three competing outcomes: tying goal, empty-net goal against, or goalie returned
• Applied a Weibull hazard framework treating outcomes as competing risks rather than collapsing them into a single event, extending prior literature on hockey goal-scoring
• Found the estimated shape parameter indicates a non-constant, accelerating hazard over time, with implications for when teams should optimally pull their goalie