The National Hockey League or the National Goalie League?
By: Dylan McCann
There aren’t many people on the planet who enjoy having frozen circles of rubber flying at their face at ninety miles per hour. If you happen to have just read that and thought it sounded fun, you may want to go buy some skates, strap on some pads, get a helmet, and sign up to play goalie for your local beer league hockey team. Yes, that is what goalies do. It sounds sick and twisted, perhaps explaining why Jason chooses to don a traditional goalie mask as he machetes his victims. Never mind that.
There’s a thought amongst hockey fans that hockey should really just be called “Goalie.” Yes, it sounds intuitively absurd to call football, “Quarterback,” or baseball, “Pitcher.” Despite this, the sentiment of “Goalie” persists regardless. This is largely due to the ostensibly disproportionate effect goalies have on the outcome of a game. While there are typically five skaters for each team on the ice at any given time, there’s only ever one goalie. A goalie does not have to forecheck, pass, shoot, take faceoffs, stick lift, poke check, backskate, and on and on and on. The goalie has one job: stop the puck from going in the net. If your goalie cannot do his job, you more than likely cannot win a hockey game. If a forward or defenseman cannot do his job, he has seventeen other teammates who can probably make up for him and win the game anyway. But you may sit here saying: “Wait! What if a team is so good defensively that they allow almost no shots? Say 2 or 3?” Well, yes that team would probably be able to overcome a goalie with a save percentage of 0. The thing is, the last time a team took zero shots in a game was in 1958. To put that into its proper perspective, Jaromir Jagr wasn’t even alive.
The point is this: an NHL season is 82 games and teams concede anywhere from twenty to forty shots a game, so you need your goalie to be able to stop the puck if you’re expecting to win even one game. This concept birthed the idea of calling hockey “Goalie.”
At first glance, this isn’t a terribly unreasonable statement. When we look at the goalies of recent Stanley Cup Winners we’re met with Jonathan Quick, Andrei Vasilevskiy, Corey Crawford, Tim Thomas, Martin Brodeur, Jordan Binnington, Matt Murray, etc… Yes, I’m including Jordan Binnington and Matt Murray because before they were throwing water bottles at people or saying shots were lucky, they were, in fact, extremely high quality goalies. As such, you’d be forgiven for seeing this list of names and their corresponding performances come playoff time and assuming that the team with the best goalie wins the Stanley Cup. It’s a fairly logical conclusion. However, as we’ve all heard time and time again, correlation is not causation
and thus we must dive deeper into the analytics of hockey teams so as to get a greater understanding of whether hockey truly should be called “Goalie.”
There are three advanced analytics which I think best demonstrate a team’s defensive, offensive, and goalie performance. A team’s defensive performance can be best documented using the stat of Expected Goals Against (xGA). xGA is essentially a stat that looks at how many shots a team gave up and the quality (location, type, situation, etc…) of each shot, and calculates how many goals we’d expect an average goalie to give up. This stat blocks goaltending in that only a team’s defensive performance will affect the xGA as it is calculated with the assumption that every team has the same quality of goaltending. Basically, if your team gives up five breakaways a game, your xGA is probably going to be higher than a team that gives up none. As for offensive performance, it’s basically the inverse of xGA. We calculate Expected Goals For (xGF) by assessing the quantity and quality of shots for. Again, we assume an average goaltender so as to isolate for strictly the opportunities that a team’s offense is generating. Having made goaltending performance the same for all teams, we can then calculate the xGF and see how teams differ in offensive performance. The best way to measure a goalie’s impact is to look at the stat Goals Saved Above Expected (GSAx). GSAx makes use of xGA by doing the following: Given we have our xGA, we can subtract our actual goals against from our xGA and determine how much better - or worse - our goalie was than an average goalie. If our GSAx is positive, this means our goalie saved more goals than we expected an average goalie to and our goalie has thus exceeded expectations. Intuitively, a higher GSAx indicates a better goalie performance.
The aforementioned analytics were compared amongst playoff teams and non-playoff teams for the NHL seasons spanning 2013-2018 and 2021. These seasons were examined because the playoff structure was different prior to 2013 and in 2019 and 2020 due to Covid. Interestingly, goalie performance has varied. From 2015-2018, it appeared as though the discrepancy in goaltending between playoff and non-playoff teams was growing, however this trend was killed in 2021 when the discrepancy noticeably decreased and became more akin to 2014 in which the quality of goaltending was not much different for lottery teams or playoff teams, relatively and generally speaking. Our offensive and defensive differences between playoff teams and non-playoff teams exhibit quite similar behaviors. In some years, goalie performance is the biggest difference between good teams and bad teams, in some years it’s offense, in some years it’s defense. It’s possible we’re seeing that offensive performance is starting to become most impactful in what’s a playoff team and what isn’t, but unfortunately this trend only spans 2017-2018 and 2021. Due to Covid and different playoff structures, we don’t have the appropriate data to examine whether this trend is consistently growing or just an anomaly much like our recent goaltending trend was. Unfortunately, these plots do not tell us much as they simply show that playoff teams typically tend to have better offensive, defensive, and goalie performance. The bigger the gap between the boxes is, the more impactful we can assume that part of hockey is. One additional interesting thing to take note of is that the tighter the boxes are, the more similar those teams’ performance in that metric are. The larger the box, the more that performance varies among teams. As I said earlier, in some seasons this is goaltending, others offense, and the remaining defense. From this data, we cannot infer that goalies have any disproportionate effect on a game and thus, hockey should remain hockey.
An interesting byproduct of this research was the finding that, since 2013, the quality of offense has gone up while the quality of defense has gone down. On the contrary, the quality of goaltending has stayed relatively consistent, exhibiting almost a sinusoidal trend. This suggests a few things, either a) Teams are getting worse at defense, b) Teams are getting worse at offense, c) Rules are now favoring offense, d) Incoming NHL forwards are better at scoring than incoming NHL defenseman are at defending, or e) All of the above. More research would have to be done to come to a conclusion on this, though I personally imagine option e is a good theory. What must be emphasized is that none of this has anything to do with goaltending. Our goaltending quality has stayed right around the same range, so even if things like save percentages or goals against averages are starting to look a lot worse than they were a decade ago, that doesn’t mean the goalies we’re watching are worse than those goalies from a decade ago. In fact, offensive performance and defensive performance appear to continue their positive and negative trends, respectively, regardless of whether goalie performance improves or declines from year to year.
After examining these plots, we predicted whether a team would be a playoff team or not according to each metric. I.e., we predicted whether a team would be a playoff team based on either their goaltending, offensive performance, or defensive performance. The table above shows the accuracy of these predictions (how many teams were accurately classified) and, again, we cannot see any clear favoritism of any metric. The goalie model is best in 2016 and 2018, the offensive model is best in 2021 and 2014, and the defensive model is best in 2013 and 2015 while tying with the goalie model in 2017. These models, however, look specifically at whether a team is a playoff team or not, thus they may not be a good indication of team quality as a bottom ranked team is much different than a team which misses the playoffs by one point.
To account for this possible shortcoming, we predicted how many points a team would garner over the course of a season according to the same metrics. Despite this adjustment, there was, yet again, no consistent winner amongst our models. 2018 and 2021 favored offense, 2015-2017 favored goaltending, and 2013-2014 favored defense. Ostensibly, there is no particular team performance or player performance metric which can be said to disproportionately influence a team’s season.
I went into this paper hoping to prove that my brother, who played goalie all his life, was the most important aspect of every team he ever played on. Unfortunately, I was not proven correct. Perhaps this is something I should have expected, though. When my brother and I lived in Australia, he was the country’s best goalie. He was so good that he was asked to play for the national team, in fact. What excellent, historical numbers did he post, you may ask? Something like 8.5 goals against per game and a save percentage around 85%. His team was not good and he definitely did not make up for that. The NHL may not be Australian youth hockey, but hockey remains hockey and is not just “goalie,” not yet at least.
Castillo, David. “The Casual Critic's Complete Guide to Hockey Analytics and Advanced Stats.”
Defending Big D, Defending Big D, 4 Oct. 2022,
Lloyd, Michael. “#EastCoastBias: Why Jonathan Quick Is Still Underrated.” Jewels From The
Crown, Jewels From The Crown, 4 Mar. 2019,
Moneypuck.com -NHL Analytics, Playoff Odds, Power Rankings, Player Stats, https://moneypuck.com/index.html.
“NHL Stats, History, Scores, Standings, Playoffs, Schedule & Records.” Hockey, https://www.hockey-reference.com/.