When Do NHL Teams Pull Their Goalie?
I’m an unusual hockey fan in a couple of ways. I’m fairly new to the sport, for one, and I also came into the hockey world analytics-first, which isn’t a particularly common experience for a sport in which the strategy of evidence-based decision-making still struggles to gain a foothold with some fans (and some teams).
That said, I was familiar with the goalie pulling argument (i.e., according to the math, teams should be fairly aggressive in pulling their goalie for a sixth skater to maximize their possible standings points) before I even really started watching hockey. I considered it fairly analogous to the fourth down in football: the math had shown that teams needed to be more aggressive in a traditionally-conservative situation, and the teams were following suit, if slowly.
So imagine my surprise when I started watching hockey and saw that goalies were being pulled with only ~60 seconds to go, maybe up to 90 seconds, when the math says they should be pulling at four minutes or even six minutes. That did not seem like much progress.
There’s lots of research and articles and takes and opinions on when teams should pull their goalie, but since I couldn’t find as much on when teams are pulling their goalie, I decided to take a comprehensive look at the state of goalie pulling in the NHL over the past four seasons. This project culminated in a dashboard, embedded at the bottom of this post (also available here), and a presentation at the Seattle Hockey Analytics Conference in April 2019 (slides available here).
A few highlights of data available in the dashboard:
Average pull times are creeping up, slowly. The MoneyPuck Pull Bot is on Twitter here.
Toronto has consistently been the most aggressive team over the past four seasons.
Pulling the goalie “works” about 15 percent of the time (when facing a one-goal deficit), and goalies tend to get pulled earlier in those situations.