2019: Binnington vs Rask - Rask is Rask, he's a good goalie. Binnington is a really interesting case. Kind of has that Craig Anderson element to him...where he's good for a 45 save shutout or he's got an early shower after giving up 4 on 16. I'm not sure if the numbers hint that at all
@Doctor No but that's my feeling. This is another good matchup though. This isn't nearly as bad as what we had to deal with early on.
This is a good question and I need to dig in on this - my gut told me that I was going to say "yes, absolutely, Binnington is super inconsistent".
And then I went to pull some data to prove it...
(interlude follows, skip if you'd like)
(Insert standard save percentage caveats and limitation statement here)
I do a calculation that compares how a goaltender's opponent-adjusted save percentage varies from game to game within a season. Just to make the example simpler, let's suppose that Goaltender A and Goaltender B both have season save percentages of 91.0%. If Goaltender A's game-by-game save percentages are 91.0%, 91.0%, 91.0%... 91.0%, and Goaltender B's game-by-game save percentages are 85.0%, 97.0%, 85.0%, 97.0%, ..., then Goaltender A is far more game-to-game consistent than Goaltender B. You know exactly what you're going to get every game from Goaltender A.
Consistency in this context can be good or bad - first of all, if a goaltender's consistently bad, that's good to know but not helpful. Second of all, especially on a bad team you might be able to live with Goaltender B's lows (since you were going to lose anyhow) but you might steal some games on Goaltender B's highs.
Fortunately there's a mathematical calculation that follows from binomial distributions that allows you to say "if I distributed the composite randomly, what would be a typical distribution?". And we can then look at individual goaltenders' distribution of game-by-game results and compare to the typical distribution.
(end interlude)
Looking at Binnington's season variances (starting at 2018-19), he's 0.95, 0.90, 0.85, 0.95, 1.02, and 0.98 (for the regular season just ending). 1.00 is "par", so Binnington is regularly more consistent than that.
To allow people to sniff test whether this method "works", here are the five most consistent goaltenders in 2023-2024 (min. 1000 shots):
Linus Ullmark (0.79)
Jacob Markstrom (0.82)
Connor Hellebuyck (0.83)
Elvis Merzlikins (0.83)
Joey Daccord (0.84)
It's important to re-emphasize that "consistent" does not necessarily mean "awesome". Merzlikins stood out to me as "hmm, I'd better double check that befoer posting" specifically here. If you look at his game logs:
One thing you'll see is that when he truly was off, he was removed pretty quickly. So on a shots-weighted basis, his stinkers were minimized.
Anyhow, the "most consistent" goaltender by this metric is Linus Ullmark. Overall, 40% of his games (shots-weighted basis) were "average", and 40%/3%/0% were +1SD/+2SD/+3SD above average. Conversely, 12%/5%/0% of his games were -1SD/-2SD/-3SD below average.
Compare against the five least consistent goaltenders:
Samuel Ersson (1.20)
Pyotr Kochetkov (1.16)
Mackenzie Blackwood (1.15)
Tristan Jarry (1.14)
Filip Gustavsson (1.13)
40% of Samuel Ersson's games were "average" - just like Ullmark, which is a nice coincidence. But 20%/8%/0% of Ersson's games were above average, and 16%/13%/3% of his games were below average. 16% of Ersson's games were at least two standard deviations below expected (compared with just 5% of Ullmark's games). And 8% of Ersson's games were at least two standard deviations above average (compared with 3% of Ullmark's). Ersson's games are a lot more spread out.
So that's a lot of words to say that Binnington's slightly more game-to-game consistent than an average NHL goaltender.
We'll include one more save percentage caveat - in particular, one thing that is likely bleeding in here would be a team's consistency in preventing high-danger shots.
