Henrik Zetterberg seems to defy Father Time. He had his best full season since 2011-12 last year at the age of 36. Without his performance in 2016-17, it’s a bit scary to think how much worse the team could have been. My two previous articles showed that even at 37, he still leads the way on the ice in many respects.
At the same time, as the saying goes, Father Time is undefeated. There will come a time where he just can’t make his body do what he wants it to do. Just like all of you, I imagine, I hope that day is far in the future. Unfortunately, it’s likely that it will be here sooner than we hope.
For this article, I’m going to take a look at two metrics to predict a reasonable expectation for Zetterberg for the 2017-18 season.
5v5 P60 and the Standard Age Curve
In 2014, Eric Tulsky wrote an article explaining his research into establishing an age curve. Most people think that NHL players peak later in their career than they actually do. Tulsky found that, in terms of scoring, forwards peak at 24 on average. The drop off is not very severe from age 24 until about 29 (roughly 90% of peak), after which it tends to drop more severely, reaching about 50% of peak at age 37. Zetterberg will be 37 this season.
There are a few things to remember here. The first is that this is an average. Some players will drop off faster, and some players will maintain a higher level of production longer. The second is that this is 5v5 points per 60 minutes of ice time, so this is predicting their rate of 5v5 scoring, not necessarily their total number of goals and assists (due to a potential change in ice time).
The latter point is important to remember because Zetterberg played a large amount of minutes last year, and it seems likely that he will be asked to do that again this season. Dylan Larkin is not ready to take over 1C duties, and unless he has a major regression this season, Zetterberg will be more able to fill that role than Frans Nielsen.
Each of the last 3 years, his 5v5 ice time has increased. It seems unlikely that it would increase much more than last year, but even if it decreased to the average of the last three seasons, he would still be averaging 14.71 5v5 minutes.
Last year, Zetterberg led the forwards in overall TOI, playing 1616 total minutes. Tomas Tatar was second with 1416, two hundred minutes behind. At 5v5, Zetterberg played 1241 minutes, while Tatar was second with 1116 (1st and 2nd among forwards again).
Even though he logged a ton of minutes the past three seasons, is it reasonable to expect that again? The older he gets, the greater the chance of injuries from that amount of ice time.
Rob Vollman, author of Hockey Abstract, has 5v5 data going back to 2006-07. In 2006-07, Zetterberg was 26, which is right about 96% of peak according to the age curve. Using this as a starting point, I overlaid the aging curve over Zetterberg’s actual 5v5 P60 for those seasons.
The age curve says that Zetterberg should score 1.46 points per 60 minutes at 5v5 next year. If you look at the overall trend of his actual numbers, you will see that his performance last year was much higher than you would expect. The two seasons prior, he was 0.5 P60 below the age curve, while last season he was 0.7 above it. To put that in perspective, that 1.2 P60 difference is a difference of 20 5v5 points a season (based on Zetterberg’s average 5v5 TOI over the last three seasons)
What we see here is a downward trend with a few outliers, last year being the largest. While we don’t want to ignore last season—he certainly earned the numbers he put up—the chance of him outperforming expectations that significantly is less than the chance of him regressing.
So by looking at this data, the likely expectation will be for him to regress closer to the age curve. But by how much?
On August 14th, Prashanth Iyer wrote an article for The Athletic (paywall) that looked at projections for the Red Wings forwards by using a Marcel projection for the 2017-18 season. Since it is paywalled, I can’t give you the full chart, but I can provide his model’s projection for Zetterberg next year, which is 1.90 5v5 P60
If you look at the chart above again, that projection would be between his performance last year and what the age curve projects. I think that 1.90 is closer to a reasonable expectation than 1.46. If he plays 82 games like he did the past two seasons and using the three year average of 5v5 TOI, that would project him to score 38 points at 5v5 down from 55 last season.
GAR and a New Age Curve
One of the reasons that I’ve been using GAR so much lately is that the model looks at multiple components instead of just scoring. So let’s take a quick look at Zetterberg’s GAR components since 2008-09.
As a reminder, DRW = penalties drawn; EVD = even-strength defense; EVO = even-strength offense; FC = faceoffs; PPO = power play offense; and TKE = penalties taken.
Each component shows Zetterberg’s value to the team in that component in terms of goals the team scored above what would be expected if you replaced Zetterberg with a replacement level player.
Every year, Zetterberg has been negative in only one category: penalties drawn. That means he draws fewer penalties than you would expect a replacement level player to draw. Each season, he provides the most value to the team through EVO. Last year he provided more EVO than the two prior years, though his PPO took a sharp decline. Because of the negative values, it can be a little tough to see how his totals add up each season, so this next chart illustrates his overall GAR for each seasons since 2008-09.
2012-13 (lockout) and 2013-14 (major injury) are pro-rated to 82 games. He played 46 and 45 games respectively in those seasons. He was performing on a higher level than normal in the 45 games he played in 13-14, but it’s impossible to tell if he could have kept that up for a full season.
Just like with P60 in the first section, we see a general downward trend, with a few outliers. Again, just like with P60, we see that last season was above what was likely for Zetterberg to produce.
Since Tulsky published his article in 2014, others have worked on age curves. Stephen Burtch, hockey Analytics writer for Sportsnet.ca, recently released GAR projections for next year based on a quadratic projection using existing GAR data and a parabolic aging curve.
Burtch made an interactive dashboard that you can use to look up any player. It doesn’t have a projection for every player, as he needs data from several seasons to calculate the estimate. Even for those players, you can still see their actual GAR totals.
In this graph, the dark-blue dotted line represents the trend for next season solely based on previous seasons. The light-blue dotted line represents Burtch’s model’s projection for next season. The model projects him at 12.92 for next season, down from 13.5 this season.
It’s important to note that the blue shaded area represents a 95% confidence interval. So, the projection is 12.92 +/ about 3.1. That’s a significant range.
Are These Reasonable Projections for Zetterberg?
I feel that yes, both Burtch’s GAR projection and Iyer’s 5v5 P60 projection are reasonable expectations for Zetterberg next season. Over time, everyone regresses. The great players are often able to keep producing at a higher level for longer than the average player, but eventually they take a step or two back. One thing I think these two stats show is just how impressive Zetterberg’s performance was last year.
The captain could defy the odds again and perform like he did last season. It is, however, more likely that he helps the team less on the ice than he did last year.
This serves to highlight how important it is that other players step up this season. Even if Zetterberg does regress, if the other forwards take a step (or two or three) forward, the team will be better as a whole.
Again, regardless of what Zetterberg is able to accomplish this season, I think it’s worth taking a moment to realize just how good he was last year at the age of 36.
Interactive charts from this article can be found here.