Is goals against average a proper way to measure a goalie’s performance?

A couple of days ago while browsing through several hockey sites, I stumbled across something on another blog that grabbed my attention: It was about Kloten’s decline and whether they still had chances to make the playoffs. The writer (or writers?) mainly oultined the departed Tommi Santala, missing depth (in the case of departed Peter Guggisberg) and Martin Gerber’s goals against average (GAA) as reasons that Kloten will miss the playoffs.

That last part is the one that hit me. I have a hard time believing that someone seriously takes GAA as a stat in order to state whether a goalie is good or bad. There’s no doubt, a high GAA costs you wins. But you can have a very good goalie behind a porous defense and his GAA is just going to suffer. So I was curious how Gerber lines up in several categories compared to his peers. In order to do some analysis, we need a proper sample size: Therefore, I just looked at goalies that have played at least 700 minutes this season. There are 16 of them in total.

Let’s get the obvious thing out of the way quickly: Gerber has the worst GAA of that lot…and by far:

 

Goalies with at least 700 minutes played, sorted by GAA
Goalies with at least 700 minutes played, sorted by GAA

Also interesting is, that Gerber and Boltshauser really do not get any support. Kloten is one of the worst possession teams in the entire league and is doing little – either systems wise or personnel wise – to support their goalie. They do not generate any scoring chances but play putrid defensively.
I applied stats per 60 minutes to get a clear overview of all the goalies and being able to measure them the same way:

 

Gerber_02

What we have now is somewhat logical: the team that faces the most shots has the worst GAA. However, that’s assuming, that every goalie is equal and stops the puck league average. Since that is not the case, the pendulum in defining if a goalie is good or not has to be another metric. And we all know that this is a goalie’s save percentage (Sv%). Over a large enough sample size, Sv% is a good indicator for a goalie’s skill level. Sorting our table by Sv% gives us the following picture:

 

Goalies with at least 700 minutes played, sorted by Sv%
Goalies with at least 700 minutes played, sorted by Sv%

Looking at Sv% allows us to do a proper assessment. What we can see now, is that Gerber is definitely not stopping the puck enough. But if Kloten were to play a bit tighter in their own end, this would already help a lot. If Kloten allowed about 3 shots less per game, that would amount to almost 0.28 goals against less per game (GAA 3.10). Each goal is worth 0.5938 points towards the standings. Which means, if Gerber still saved with his .908 Sv% but had faced only 745 instead of 814 shots on goal, Kloten would have received about 6 to 7 goals less. This does not sound like a lot, but it’s actually worth between 3.5 and 4 points in the standings. And this is worth a lot.

Getting back to the beginning where I said that GAA is not an appropriate metric to measure a goalie’s performance. A good example in this table is Damiano Ciaccio. He faces the second most shots per 60 minutes. He has the sixth highest GAA and most importantly: The fifth highest Sv%. Looking just at GAA, we would assume that Ciaccio is another poor goalie. But taking a closer look at SOG A60 and Sv% shows that he is actually doing everything he can to keep his team above water and all they do is just…well…nothing. Like Kloten, Langnau is just a poorly run team that would need some adjustments to make life easier for their goalie.

To summarize: GAA shouldn’t be used to assess a goalie’s skill level. It’s a stat that is highly influenced by the skaters in front of him and therefore by the shot volume he faces. If Kloten really wants to get into the dance and make the playoffs, they should consider changing their defensive system and playing a bit tighter so they do not allow as many shots. Or quite simply: Just put the other guy in the net that actually stops a few more pucks AND adjust your system. Then you don’t have to go into mid-January wondering whether you might have to play one of the NLB’s powerhouses in a few weeks.

All stats were put together as of January 19
 
Another version of this post is published on SwissHockeyBlogs.ch