By The Numbers: The Jets best defensive seasons

The Winnipeg Jets 2.0 have been around now for five full seasons (well… four full and one partial, lockout-shortened, season). This gives us ample data to construct some fun lists of the best and worst statistical seasons for performers.

We take a look at some of the best and worst statistical performances by Winnipeg Jets’ defenders.

A Primer on GAR

What is GAR?

Fans of baseball or basketball should already be familiar with the statistic WAR, or Wins Above Replacement. WAR combines the many output measures of a player into one currency, making it easy to compare between two individuals in their overall performance or understanding the relative importance in different aspects of the game.

GAR, or Goals Above Replacement, combines the different outputs of a player and estimates their total impact on a team’s goal differential, relative to the estimated impact of a replacement level hockey player. The number of goals it takes to equate a W in the win column differs on the year.

The statistic looks at things such as shot differential, shot quality, faceoff, special teams, and scoring performances and then compares them relative to their ability in garnering and predicting wins.

It should be emphasized that the purpose behind GAR is not to be a be-all, one number to rule them all. But rather, it should be a starting point, with studying each component and microstatistics afterward in player evaluation.

DTM About Heart, of Hockey-Graphs, constructed the GAR we will be using, and his posts on the construction of the metric should be out sometime soon. I will note that I am excluding penalty differentials for this series, which matters (as Jets fans should know more than most) but I excluding due to some huge outliers in the data and more research being needed in area.

Jets Defensive GAR

GAR

2012-13 numbers have been extrapolated to estimate a full season’s value.

A few members on the list may stick out as a surprise, but generally speaking we see the Jets top players on the left, and their depth players on the right.

The one issue with looking at a GAR as an aggregate statistic is that usage is not always optimal. Players get hurt due to freak incidents and sometimes a coach can over play the less effective player. So, to account for this I then broke down the Jets’ GAR relative to their ice time.

GAR60

Not much changes, which should be expected unless there are HUGE issues with coaching choices, but there are some changes.

I was originally going to do these series as a top/bottom-performers, list-type article, but unfortunately it becomes boring quickly as we see the same names constantly in both cases (and the same is true with the Jets’ forwards).

Instead, I’ll just go over some points.

Notes:

  • Dustin Byfuglien is really, really good. Not one Jet has a single season as good as his “second worst” season, which is a pretty incredible accomplishment. A quick surface level scan I made seemed to indicate that a GAR/60 of over 0.4 seems to be Noris-talk worthy.
  • Byfuglien has been trending downward over the past three season, but on the positive side he has a long way to go down before he becomes a normal human being, let alone a liability. While Byfuglien is at the age that a downward trend is expected, I wonder if his recent commitment to fitness has been the cause to the decreases being so marginal.
  • More on Byfuglien and stats, people often say that stats ignore the defensive lapses Byfuglien often has in his own zone, which seems to be completely wrong. Interestingly, Byfuglien actually has three of his five seasons being below replacement level in defensive impact (and four are below any of Stuart’s!). His elite results come from the fact that his impact on offense is just so much greater than his impact on defense.
  • Toby Enstrom is just underrated. Even at his age, and despite not being very recognized, he puts up great numbers as a legitimate top-four defender.
  • While Byfuglien had some of the worst defensive impact seasons in GAR, Enstrom had some of the best. His offensive impact was above average, but Enstrom out performs individuals in defense more than most give him credit for.
  • Why I think Enstrom is so underrated is that people misunderstand defensive impact. While most people rate defensive impact by play in the defensive zone, true defensive impact is actually the absence of things happening in the defensive zone. It’s like how people used to say the opposition dumps the puck on Enstrom because they are trying to expose his size, when actually it is because Enstrom’s gap forces the uncontrolled zone entry.
  • Speaking of defensive impact, Hainsey and Pardy both were ineffective offensive players, but were positive contributors to the team through their defensive play. Jets’ fans are constantly hoping for a defensive left-shot defender, but when they have them many tend to hate them.
  • Jacob Trouba, at least on the ice, will be missed and will be incredibly difficult for the Jets to get equivalent value in a trade. He has not only been one of the Jets best GAR defensive performers at a young age, but one of the best performers in the NHL at his age.
  • While Enstrom on average had the best defensive impact, the very best defensive season was actually Trouba’s 2015-2016 season. Trouba is basically the best offensive performer after Byfuglien and with the defensive performance of Enstrom.
  • Clitsome was the biggest surprise for me, as I always thought his numbers were comparable or better than Chiarot’s. His first year was strong though, so I wonder how much of his steep decline was caused by the severe back injuries.
  • Speaking of the Chiarot-Clitsome, if early play this season is any indication, Chiarot may actually be following a similar steep decline, although without the potential back injury explanation.
  • Postma comes out as a solid depth defender. While he is flawed, and I would hazard he would be exposed more if he consistently took big minutes, Postma plays like one would hope to get out of their third pair defender.
  • The Jets pulled a fast one on the Buffalo Sabres with the Myers to Bogosian swap. Myers may not be much better than what Bogosian was early on, but he is better than what Bogosian has been lately. This is all while Myers’ salary falls below his AAV, when Bogosian’s does the opposite.
  • Bogosian actually may be what many thought Byfuglien is. He has decent offensive numbers, but not strong enough to significantly overcome his poor impact in defensive impact. Bogosian’s defensive impact was even worse than Byfuglien (the only one aside from Harrison and Ellerby who was), and without the elite offensive impact.
  • The last player I want to touch on is Mark Stuart. It is amazing how consistently Stuart has been below replacement level, despite having (barely) above replacement level defensive impact. Stuart may not hurt the Jets in the goals against (well, he does, but not more than a replacement level player would, at least) but he does hurt the team overall in the wins column, and more so per minute than any other Jet.

For your enjoyment, here are the total GAR/60 for each defender over their duration in Winnipeg:

GARcareer

The above are just snap shots of averages. Players with small samples can have outliers that skews the mean for individuals. It also doesn’t show any direction or projection of future performance either. For example, Pardy has one season superior to Myers and one season below. Averages make out Pardy to be better, yet it was the most recent year where he struggled and his struggles caused him to be scratched more (leading to the weaker year having a smaller impact on the average than the stronger year).

Closing thoughts

One must remember that GAR is not a perfect measure.

Just like in all performance, the smaller the sample a player is looked at, the more likely it is that outlier performances skew the results from the player’s “true talent” level, as we just spoke about.

The model is also still in early developmental stages. While we are at a Minimal Viable Product, there is still likely research that will influence the weighting of the components that go into the model. While the number in general seems to match my eyetest and the results come out as I would mostly suspect, that does not mean minor adjustments could change a few rankings here or there.

One-number statistics are of course the starting point. It gives us a general idea how each player has performed in relation to others, despite differences in skill sets. However, performances are not exclusively by individuals and their skillsets, but also how these individuals perform in the particular environment. Different environments, like coaches, systems, chemistry, and psychological factors, could cause a slightly weaker GAR player fit better than the slightly beater GAR player in the situation.

GAR also does not tell you how best to deploy your players. Nor does it tell you which areas a player needs improvement upon for developmental purposes.

What it can do is combine the performance statistics into one currency and give you a handle of how different players have performed for the Jets.

The Jets have carried some strong defenders like Byfuglien and Enstrom, while they have some promising players like Trouba (well… depending on certain things). They have also carried some large drags for far too long, like Stuart.

Overall, GAR gives us another way to look at the Jets and their players and also see who has performed their best.

  • FishWhiskey

    Nice work Garret! I hope the Jets are pushing some work your way. Having your analytical expertise on staff could save them years of agonizing over whether to keep liabilities like Stuart on the team at the expense of demonstrably better talent.

    Jets management have made some good moves over the last five years but they have balanced it out with a few real boners. Every one of their bad moves could have been avoided if they had made use of analytical expertise to fact check their gut feelings. A lot of “old school” hockey guys don’t like having their judgement challenged with silly things like “facts”. Real professionals, however, put their egos aside and make use of every available resource.

    If the Jets are to make the most of the young players they have drafted they need to move into the 21st century and inform their decisions with hard data.

  • FishWhiskey

    While on the topic of defence……What’s up with Big Buff?

    I can’t recall him throwing a single devastating hit yet this season. He seems to be playing a very Lady Bing like style of hockey and is making some nice fineness plays but none of the bone crushing statements for which he has become famous and that do so much to energise the team.

    Every time the TV camera cuts to him in games he seems to have a moopy look on his face like Mr. Buffleupagus. Something seems to be eating the big guy. It’s bad enough to lose Little to injury but to have Byfuglien skating around out there like a forlorn calf is devastating to this young team.

    Come back big buddy, the team needs you!

    • Eddie O rules!

      I thought I remembered seeing something somewhere that Buff’s best numbers came when he played around 22-23 minutes. Do have any insight on that? I just remember that other times when he’s been playing up around 30 he wasn’t nearly as effective as when the Jets kept him under 25…

      • Well, he hasn’t been as effective this season and he’s been logging big minutes…

        That said, I’m not sure that is true. Even if it were, I think the data would likely show you the opposite. There is a lot of circumstantial factors to how much Byfuglien plays.

        If the Jets are trailing, they are likely going to play him more to catch up as their greatest offensive weapon.

        Also, who else is playing? Are they injured? Inexperienced? Are they worse than the other alternatives or better?

        In addition, you likely find that Byfuglien’s icetime depends on how he and the other defenders are playing. If Myers struggles, it probably goes up. Same if Buff were playing well. Opposite would happen for the opposites.