Judging goalies is really hard, and my suggestion is you don’t try unless you’re willing to look foolish a lot. I was not a believer in John Gibson at the time Frederik Andersen was traded to the Leafs. I thought he was AHL-ish in the World Cup of Hockey, and I continued to doubt him for a time.
I was really wrong on that.
I did totally believe in Andersen, which is a consolation. The Ducks made the right choice for them, and the Leafs made an excellent trade. When does that ever happen?
One reason it’s hard to judge goalies is that the information from the AHL or junior leagues is terrible. All-Situations Save % is so swamped in team effects that when you’re trying to make a fine measure, you’re using a ruler printed on elastic to do it with. Not only that, the entire leagues themselves vary much more widely than does the NHL, so you don’t necessarily get similar quality of shots faced across teams.
Once goalies get in the NHL, life is easier on armchair analysts. You don’t have to scout every goalie visually for hours to form an impression, and frankly, knowing what to look for in doing that is a skill few people have. “He looks calm” is the level you get on television, and I’ve spent years trying to understand the position, and I’ve barely scratched the surface. But in the NHL there is more information, better data, and lots of it.
When I looked at how many good goalies their are, with an eye to assessing this tandem idea that everyone is confidently talking about, I posted this:
This gives you an idea of the spread of talent in each year, but it names no names. Let’s do that and see if the good, the bad and the replacement-level are who we think they are.
I went to Evolving Hockey and collected up the cumulative all-situations, score-adjusted results for all goalies who have at least 30 games played and 300 Fenwick against from the beginning of 2014 to November 12 of this year. For some goalies, it’s massive numbers of minutes, and for a few, it’s a little bit.
EH included a few stats on their goalie page that need to be defined first.
Sv%: this is the traditional percentage of Shots on Goal that don’t become goals measure
FSv%: Fenwick Save % is the percentage of all unblocked shots that either miss or are saved
xFSv%: Expected Fenwick Save %: Expected Goals models are built on Fenwick (unblocked shots), so this is the Fenwick Save % a league average goalie would have had under the Expected Goals model
dFSv%: This is the difference of the Fenwick Save % over Expected.
GSAx: This is the total goals saved over Expected.
Okay, onto the fun. It’s 95 players, so if you want a Google Sheets Format: You can view it here. You should be able to sort or filter that version, and this one won’t be useful on AMP pages at all.
Historical goalie performance 2013-present
The table starts out sorted by Delta Fenwick Save %. The reason I use this measure is that it takes enough of the team defensive quality out of the picture to give you a better look at the goalie’s actual performance.
Save % can show you that Gibson is better than Niklas Backstrom without any help. It can show you that Marc-Andre Fleury is better than Mike Condon, too, but as you look at the players in the middle, you’ll start to see that more and more of them look better by one measure over the other to the point that the rankings lose their agreement almost totally. The closer you are to average, the more a team being generally good or bad, giving the goalie a better or worse chance to succeed shows up in simple Save %.
Look at Andrei Vasilevskiy, last year’s Vezina winner, and you see a goalie whose career is below league average. Some of that is that he was developed in the NHL and wasn’t immediately great, and he’s young yet, and will show better with more mature seasons under his belt. But like Ben Bishop before him, he’s helped immensely by the quality of the shots the Lightning allow. Sort the list by Expected Save %, and see who has an easy life.
The Leafs can’t have a goalie that needs an easy life. They need Freddie. And when it’s time to go backup shopping, like, oh, a few months back, they need a backup who can handle a team that is not built for defence, but manages to be okay enough at it.
You likely noticed one name near the top of this list at first. Andrew Hammond. This isn’t a perfect way to produce this dataset, but he’s a nice lesson in not falling for short bits of results over multiple seasons strung together. Remember when that lesson was Connor Carrick? Did we learn it? Maybe not, but Hammond has been played enough in the NHL on his hot streaks that his overall numbers are misleadingly high. It’s likely Garret Sparks is lower than he should be for that reason too, but not to an extreme degree, I don’t think.
I included the GSAx column just so you can see what a long career can mean in impact. Or, in Gibson’s case, how facing Carlyle hockey for years while being elite can make you the best man on this list by that measure. The Leafs are busy doing that for Andersen too. He’ll be jostling Gibson for the top spot in no time.
That’s actually a drawback of using the Delta Fenwick measure. It won’t tell you who can handle a heavy workload, and who can’t. Gibson, Andersen, Crawford and Halak haven’t needed as many games played to top that ranking as some of the others.
Two really interesting names here are Alex Stalock and Keith Kinkaid. I don’t understand why either one has lasted as long as they have, but sometimes a team plays the only goalie in the system who is available. If you help Stalock out a lot with shot quality, he can look league average, though. The Leafs were right, in hindsight, not to ask him to stay when they ended up with him in a trade.
Okay, now it’s your turn: go forth and argue about goalies.