February 10 2017 03:00PM
When the Canucks pick up two points this season, Ryan Miller is the story more often than not, much like he was last night in a 3-0 shutout victory over the Columbus Blue Jackets, in which he was the first star. With a goals per game of 2.30, better than only Arizona and Colorado, the Canucks often need to rely on goaltending to squeak out a victory. That was certainly the case after Christmas and through much of January, when Miller was the league's hottest goaltender and lo and behold, the Canucks manufactured a six game winning streak.
But then, as Willie Desjardins continued to ride his number one netminder, Miller's numbers began to erode. Damn near everyone, from fans to analysts to the Canucks staff themselves, has pointed out that Ryan Miller plays better when rested - and I have the numbers to prove it.
So why are they still running him down?
Last season, Ryan Miller was clearly the Canucks number one goaltender, and he was played like one for much of the season. Right up until it was clear that their season no longer mattered. From that point on, Willie split Miller and Jacob Markstrom's starts, alternating them back and forth until season end. The result? Reliable goaltending.
So we said, going in to this season, why not do it again? This is a club that was desperate to show that it has rounded the corner and is now on the upswing. With the roster they'd assembled, they were going to need all the help they could get, and that included great goaltending.
So an even rotation seemed like a win for everyone. Ryan Miller stays fresher. Jacob Markstrom, the goalie the Canucks just doled out an $11 million contract to, continues to develop. And the Canucks get a better chance to win games on a regular basis. Win-win-win.
But then Jim Benning came out and said this:
#Canucks Benning on 1040: 'Ryan Miller is our number one goalie. Let's make no mistake about that.'— Jeff Paterson (@patersonjeff) September 19, 2016
That seemed like an awfully bold move considering all of the points I just made.
As the season has worn on, Ryan Miller has had his hot and cold streaks, as has Markstrom. While Miller was getting a lot more credit for getting the Canucks near a Wild Card spot (back when they were near a Wild Card spot), the two goalies' numbers are pretty similar. It's just that Miller is getting more playing time, especially of late.
As of last night, Ryan Miller has made 13 of 19 starts for the Canucks, more than doubling Markstrom, who has made just six starts over a six week period - not exactly the grooming for the starter's job that we expected to happen.
As with many areas of many sports, coaches have a tendency to ride the hot hand when it comes to goaltending. Willie Desjardins has certainly done that with Ryan Miller during their respective tenures with the Canucks - despite the fact that there has been ample evidence that Miller eventually falters when he plays too much.
Speaking of evidence, I've compiled some that demonstrates this very point.
My hypothesis: that Ryan Miller get worse under heavier workloads. The test for this is simple: I compiled every game Ryan Miller has played as a Canuck (via Corsica.hockey), then labeled each game according to how many appearances Miller had made in the previous 5, 10, 15, and 20 days. I then calculated the average save percentages (a variety of different ways) at each level of prior start frequency, and then let my graphs get to work.
Here are Miller's save percentages, as well as Corsi-save percentages and Fenwick save-percentages (all at 5-on-5, so varying frequency of special teams play doesn't muddy the data) over the frequencies of appearances in the 10 days prior to a start:
As you can see, the downward trend is very pronounced. While one appearance (in 10 days prior to a start) is better than zero (likely a function of regaining timing after being out of game action), there is a very steady decrease after that. This graph also contains the league averages (from 2014-15 to present) in each category, denoted by dotted lines.
It should already be clear that Miller performs better under a lighter workload, hovering above league average only at the one and two appearances mark, but I'll provide more evidence anyways, just to drive the point home.
The next graph show three other types of save percentages, on low-, medium-, and high-danger shots against, plotted against the same frequencies of appearances, again with league averages denoted by dotted lines:
This graph gives us a slightly different perspective, breaking up the difficulty in shots against, but the picture that it paints is even more interesting. Note that his low-danger save percentage (green) stays roughly stable, while his medium-danger save percentage (blue) shows a slight decrease as workload increases. However, it's the high-danger save percentage (yellow) that really tells the tale: after a huge increase between zero and one appearances, it takes nearly a 15% between one and four appearances. That is, quite frankly, a staggering drop.
To illustrate how this affects Miller's numbers on the whole, we'll look at the proportion of the shot-danger levels that he faces. It should be fairly intuitive that goalies face fewer high-danger shots than any other type, while the other two types appear in roughly equal quantity. Let's look at the breakdown Miller has faced in two and half seasons in Vancouver:
The fact that Miller's low-danger save percentage stays stable, combined with the fact that low-danger is the most frequent type of shot faced, is certainly a boon for his save percentage. However, the devastating drop in high-danger save percentage more than outweighs the stability of the low-danger save percentage, leading to a precipitous drop in overall save-percentages.
As a result, despite facing about twice as many low-danger shots as high-danger shots, Miller has allowed over four times as many high-danger goals as low-danger goals.
really seems like it should have been dealt with long ago, but instead it has
gone on to the point where it has likely cost the Canucks several points in
the standings - some that would help them achieve their stated goal of making
the playoffs - as Miller as started roughly half of his games in the past two
and half seasons under workloads in which he struggles. That needs to stop.
The solution: the Canucks should just do what they did last season and alternate Miller and Markstrom for the remainder of the season. It affords the Canucks the multiple benefits that were talked about at the beginning of this article: Ryan Miller plays better with rest, Jacob Markstrom gets development time, and the Canucks get better all round goaltending.
At this point, the two netminders' numbers are pretty similar:
Yes, there is a half a percent difference in save percentage. Over the course of an 82-game season, given the Canucks shots-on-net against per game of 31.1, that half a percent equals about 12.75 goals, or 0.16 goals per game.
Now imagine two scenarios - one where Ryan Miller makes 48 starts and Jacob Markstrom makes 34 starts, and another where the goalies split the starts evenly, playing 41 games apiece. In this scenario, the Canucks face an average of 31.1 shots-on-net against per game regardless of which goalie is playing, and they maintain their current respective save percentages. Here's what we get:
As you can see, that half a percent difference in save percentage between Miller and Markstrom accounts for ONE goal when going from scenario one to scenario two. And that is before taking into account the differences in save percentages due to workload. It's quite likely that Miller's save percentage would be better with more rest. It's also quite likely that Markstrom's numbers would improve, as a 27-year old goalie who is currently playing about once a week.
I've taken the liberty of charting the workload for the remainder of the 2016-17 schedule for Miller, assuming alternating starts, and assuming Markstrom plays Saturday morning and Miller takes Sunday's against Buffalo, his former team.
Notice how he rides a nice, easy workload for the remainder of the season, one
with which he can stay rested and, as a result, play better. This would be an ideal scenario, and one that would be incredibly easy to follow. As of now, it doesn't seem like the Canucks are planning too far ahead.
Desjardins says no decision has been made yet on goalies for both games this weekend, but Miller likely starts tomorrow vs Bruins.— Vancouver Canucks (@Canucks) February 10, 2017
Let's just hope they've learned something from their past mistakes and don't head down this road:
Miller looks to be starting vs Boston.. Either taking a start in Buffalo away from him, or giving him 3 gms in 4 days. Hopefully the former— Blake Price (@BlakePriceTSN) February 10, 2017
This is incredibly simple, but because it's been stated so many times and hasn't been put into practice yet, I'm going to use really big font to get my point across.
STOP OVERPLAYING RYAN MILLER
Now you've seen the numbers, so there are no more reasons to continue with this travesty. Let Miller get his rest - the team will be better for it.