Is it possible that Frederik Andersen has been the best goalie in the division so far? Anything is possible, of course, but that seems unlikely after last year. How do we figure this out, though?

Many people like all-situations save percentage as the final word on goalie quality. We can start there today, and see if that tells us anything. This is Fenwick Save % or, the percentage of unblocked shots that are not goals.

  1. Mike Smith - 2 GP - 99.15%
  2. Jake Allen - 6 GP - 96.09%
  3. Jack Campbell - 2 GP - 95.96%
  4. Laurent Brossoit - 3GP - 95.48%
  5. Connor Hellebuyck - 10 GP - 95.48%
  6. Jacob Markstrom - 11 GP - 95.41%
  7. Frederik Andersen - 12 GP - 94.95%
  8. Carey Price - 8 GP - 94.75%
  9. Mikko Koskinen - 13 GP - 94.09%
  10. Matt Murray - 12 GP - 93.72%
  11. Thatcher Demko - 9 GP - 93.62%
  12. Braden Holtby - 8 GP - 93.51%
  13. David Rittich - 2 GP - 92.31%
  14. Stuart Skinner - 1 GP - 92.31%
  15. Marcus Hogberg - 7 GP - 90.87%

Only one team, Edmonton, has played three goalies so far, so maybe all the talk of the importance of the third goalie and the Taxi Squad has been a little over-hyped. To get some context for this list, if you’re not used to Fenwick Save % where the numbers are bigger, Mike Smith is the second best goalie in the NHL right now behind a fellow I’ve never heard of who got a shutout in San Jose, Alexei Melnichuk.

A lot of these guys are backups with too few games to take seriously, so let’s see the same list with only those who have six or more games.

  1. Jake Allen - 6 GP - 96.09%
  2. Connor Hellebuyck - 10 GP - 95.48%
  3. Jacob Markstrom - 11 GP - 95.41%
  4. Frederik Andersen - 12 GP - 94.95%
  5. Carey Price - 8 GP - 94.75%
  6. Mikko Koskinen - 13 GP - 94.09%
  7. Matt Murray - 12 GP - 93.72%
  8. Thatcher Demko - 9 GP - 93.62%
  9. Braden Holtby - 8 GP - 93.51%
  10. Marcus Hogberg - 7 GP - 90.87%

Now that looks a lot more plausible, with only a few teams playing anything like a tandem (Montreal should keep it up) and several teams with an obvious starter who plays most of the time. Connor Hellebuyck, who on career record, has to be the best goalie in the North is right on top. Sorry Markstrom fans, he’s had too little track record to take that title. He has been good, and so has Andersen. The rest are their team’s problem, not the solution, with Carey Price sitting in the grey area of likely good enough for his team, but not really helping them.

Notice that the Ottawa and Vancouver goalies are at the bottom, just like those teams are in the standings. So which is the chicken and which the egg? Does the bad team make the goalie bad or does the bad goalie make the team bad? Yes, is my answer to that one.

To get the sense of how the team and the goalie exist in their feedback loop, we can look at Expected Save % and the difference between reality and Expected.

I’m using Moneypuck’s numbers because (and this is how I always decide whose xG model to use) their web interface for Goalie information makes this exercise easier. Moneypuck’s xG is more exuberant for power play shots, and you will see differences between it and others, but if you’re looking at a list like this as a definitive ranking or studying the second decimal place of a percentage to decide who is better, you are, absolutely, doing it wrong.

Expected Save % is a calculated number based on the xG model for the shots taken that says what a league average goalie would do facing that workload. And no, it does not include pre-shot movement of the puck, who is shooting the puck, the goalie’s movement, the phase of the moon or any other yabbut complaint designed to draw attention to the lack of perfection in the model. This is an average that gains meaning the larger the number of shots becomes. Smaller samples are always subject to larger variance.

Today is about getting a hint of who has been good while considering his team, without trying to consider the team in our heads where we’ll be as accurate as any gut feeling ever is. This is the same 10 goalies, only now the numbers are Expected Save % and the delta (the triangle) between expected and actual. The order is by delta.

  1. Jake Allen - 6 GP - xSave: 94.41% - ∆: 1.684%
  2. Connor Hellebuyck - 10 GP - xSave: 94.45% - ∆: 1.034%
  3. Frederik Andersen - 12 GP - xSave: 94.45% - ∆: 0.502%
  4. Jacob Markstrom - 11 GP - xSave: 95.07% - ∆: 0.342%
  5. Carey Price - 8 GP - xSave: 95.32% - ∆: -0.566%
  6. Braden Holtby - 8 GP - xSave: 94.27% - ∆: -0.762%
  7. Thatcher Demko - 9 GP - xSave: 94.47% - ∆: -0.849%
  8. Mikko Koskinen - 13 GP - xSave: 95.15% - ∆: -1.058%
  9. Matt Murray - 12 GP - xSave: 95.28% - ∆: -1.529%
  10. Marcus Hogberg - 7GP - xSave: 94.67% - ∆: -3.798%

One thing that jumps out at me is that the Expected numbers are not hugely disparate. Carey Price and Matt Murray are virtually identical in difficulty of their jobs, and they lead this 10-goalie list by Expected Save%. Koskinen and Markstrom are right in there too, in what is the middle of the pack in the NHL. They call the North Division the “defence optional” division for a reason. No team is actually good at limiting the overall quality and quantity of shots against, but those four teams are mediocre league wide in how hard their starters’ lives are.

Demko, Andersen and Hellebuyck are all virtually tied, with Allen and Holtby close. The technical term for their situation is “shelled repeatedly by opponents invited to skate up the slot at will”. The goalie facing the worst Expected Save % league-wide is Andrei Vasilevskiy, and he also has the sixth best actual Save % of goalies with 6 games played or more. He wants a second Stanley Cup, even if he has to earn it himself.

Of these North Division goalies, Price and Andersen face the fewest shot per game, but Toronto allows a lot more high-quality shots than Montreal does. Vancouver and Ottawa have allowed very large volumes of shots, so their goalies get shelled with sheer numbers.

Who is the best goalie in the North? Not Jake Allen, that’s not enough games to judge from, and I only set the minimum so low because I wanted to laugh at the Price price. He is a $10.5 million league-average goalie.

Hellebucyk wins for me because he has a long history of excellence backing his early, well, excellence this season. He is elevating Winnipeg, who are a little better than expected, firmly into a playoff spot. Markstrom is next, and Markstrom and Andersen are essentially doing the same job. They’re papering over the defensive flaws in teams who think offence first. I’ll tie them for second.

After that, the goalie is part of the problem. Price could trend back up, but the rest of them aren’t likely to. Marcus Hogberg is not an NHL goalie, and should not play until he’s had about 25 quality starts in the AHL. Matt Murray is not as bad as he looks, but he’s been bad. Vancouver is so bad, they’re making some not horrible goaltending look worse, and that team is doomed this season. Edmonton should fire their GM, because he’s costing them a guaranteed playoff spot with an inexplicable set of goalie choices.

Using the Expected Goals information on top of the Save % provided some contextual shading. Largely it illustrates how mediocre Price has been. We were able to see the line where good enough becomes a liability. It’s not easy being a goalie in the North Division, but the fact only three of them are really succeeding at it is one reason why the standings look the way they do.

Analysis