The Red Wings are on track for a historically bad season. Jimmy Howard’s play has seemingly fallen off a cliff. Both of those things seem obviously true from anyone even paying a bit of attention to Detroit this season.

I wanted to take a look at just how much Howard has fallen off, and how much can be blamed on him as opposed to a poor team defense. Also, how reasonable is it to assume that he can regain some of his previous form?

Jimmy Howard’s Season

For this article, you are going to see two individual goalie statistics. I’m using these because they are much better evaluative tools than something like unadjusted Save Percentage or Goals Against Average.

The first statistic is Goals Saved Above Expected taken from the Evolving Wild Site. I’m choosing this because it’s a fairly simple statistic to understand and it partially takes the defense of the team in front of the goalie into account. It’s simply Expected Goals Against minus actual Goals Against. The more dangerous the chances allowed, the higher the Expected Goals Against will be.

I looked at goalies for this season with a minimum of 250 Fenwick (unblocked shots) against. This cut off at 62 goalies, or 2 per team.

First we’ll look at a scatterplot with unblocked shots against on the x-axis and GSAx on the y-axis. While this article mainly focuses on Howard, I included Jonathan Bernier for reference. The vertical line is the average number of unblocked shots goalies have faced.

Howard is below league average in the amount of unblocked shots faced, and he is dead last in Goals Saved Above Expected (-16.6).

I wanted to show shots faced because since GSAx is a cumulative stat, the fact that he’s racked up the worst total in the league while facing fewer than the league average in shots faced shows the extent of the problem.

Looking at this as a bar chart, you can see just how poor Howard’s performance has been:

The other statistic I looked at is Goalie Goals Above Replacement (GAR) , also from the Evolving Wild site.

Once again, Howard is dead last in the league.

The scary things for fans of Detroit and/or Jimmy Howard is that it seems like his performance has declined rapidly. Prashanth Iyer looked into this, and found that there is a good reason to feel that way. It has.

While Howard has been up and down within seasons for his entire career, this season shows a drastic downward trajectory in his Goals Saved Above Expected.

How Much of it is The Team?

First of all, Bernier’s numbers are noticeably better than Howard, and he’s playing behind the same defense. Additionally, both of the statistics used are cumulative, so Bernier has produced better numbers while facing more shots.

Still, let’s take a look at the amount of shots the team is letting up, as well as the quality of those shots.

First up is Even Strength Unblocked Shots Against per 60, once again taken from Evolving Hockey.

The higher the number, the worse, since you are giving up more shots. Even though Detroit is not doing great, there are seven teams ahead of them, and you can quickly see that it’s not that the team is giving up many more shots than the rest of the league.

So that’s the quantity of shots against, let’s look at the quality of shots against. For that, we are using Even Strength Expected Goals Against / 60 minutes (xGA60)

This time, Detroit is closer to the bottom, so the quality of shots they are giving up is worse relative to the rest of the league than the quantity. Even so, Detroit’s 2.56 is not drastically different from the league average of 2.49.

It’s important to remember that FA60 affects xGA60. Adding more shots against increases xGA60, although the higher the quality of each shot, the more the former increases the latter.

Say for example, the team gives up 100 shots that each have a 0.01 xG value. That means that they have a 1% chance of becoming a goal. That’s the same increase in xGA that you would get from giving up 10 shots that each have a 0.1 xG value. Detroit being “better” in FA60 than xGA60 tells me that the issue is more the quality of the shots they are giving up than the quantity, although obviously cutting down on the quantity would help too.

Basically, Howard’s numbers are so low that it seems very hard to blame the team defense for much of it.

So What Next?

One of the projects I’ve been working on is an aging curve for goalies. For anyone not familiar with an aging curve, it estimates the average rate of decline for players as they progress through their career. Importantly, it is important to remember that it is an average. Some players will decline more slowly, some will decline more rapidly.

Also, before debuting version one of my goalie aging curve, I want to emphasize that I think there is still work to do. Please don’t use this as anything definitive. I do think that in its current form, it shows the basic shape that a more refined version will.

The current version uses the Delta method that the Evolving Wild twins used in their aging curve, the first article of which can be found here.

Basically all seasons for every goalie in the dataset played are grouped by how old the goalie was the majority of the season. The average change between a player’s 19 year old season and 18 year old season is calculated, and so on. The next step is to use that information to track the cumulative change from year-pair to year-pair.

I used the Evolving Hockey Goals Above Replacement as the metric. This covers the seasons from 2007-08 to 2018-19. The 2012-13 lockout season is pro-rated.

One obvious aspect that jumps out is that the two ends have groups that only have a few players. In fact, there is only one goalie who played in both his 19 year old season and his 20 year old season (Robin Lehner).

In the 38-39 group, since there are only six players, one player having a major positive change is enough to create a brief uptick before the decline continues. Dwayne Roloson jumped over 20 GAR from his 38 to 39 season. If you are curious, here are the 6 goalies in that group (in order of their GAR during their 39 year-old season:

So after all that, here’s version one of a goalie aging curve.

This curve has an average peak from age 25 to age 29, which is later than skaters. From there, the trend is downward, starting to decline more steeply at age 34.

As said above, I plan to continue refining this. One possible refinement would be to account for survivorship bias. I also plan on looking at using regression like CJ Turtoro did in his recent work on aging curves for skaters.

Conclusion

Jimmy Howard is in his 35 year-old season, and on average, he can be expected to continue declining. Since his rate of decline has been extreme from last year to this year, if he continues to play following this season, it would make sense to expect that he would bounce back somewhat. The issue is: Do you want to take the chance that he can bounce back to at least be a capable backup as opposed to the possibility that he could just be done as an NHL goaltender?

As a fan, I want to believe that he can re-gain some of his form. I’ve always liked him, and I think he’s been underrated over the course of his career. From an objective standpoint, I wouldn’t want to bet on it.