The Devils were coming into this season with a truly embarrassing defensive corps. It was likely to be composed of some combination of Andy Greene, Steve Santini, Damon Severson, John Moore, Ben Lovejoy, Mirco Mueller, and Dalton Prout. This would likely be the worst group of defenders in the NHL. That group got slightly less worse when the Devil’s uncharacteristically won the sweepstakes for Hobey Baker award-winner, Will Butcher.
He was expected to compete for a job right away, and, with good performance in the preseason, solidified his role as a top PP defender and 3rd pairing guy at 5-of-5 play. The rest was history. This year, Butcher has been, debatably, the most productive defenceman on the team and, non-debatably the most efficient one. Yet, he continues to get among the least amount of time (typically last among defenders). In this article, I’ll briefly show how he is performing in several classes of statistics and comment on if/why he should be given more opportunity as well as speculate as to the reasons he may be being sheltered from it. All charts below are 5v5 from Corsica and the relative stats are adjusted using Manny’s global adjustment which includes score, venue, and some other things.
Will Butcher got a lot of secondary points and powerplay points early on that I think people dismissed some of it as noisy or fluky. Well, at 5v5, he is the most productive D-man in total points, and trails only Severson and Santini (? ...okay whatever) in primary point rate (P1/60). This is unsurprising given that he is the most prolific shot generator in games tracked by Corey Sznajder. He is generating he most chances, and predictably getting the most results as a consequence.
This is only 5v5 production. In the interest of pretending to be unbiased here, I excluded powerplay points — with which he becomes the 3rd highest scorer on the team and the 21st highest scoring defender in the NHL (after last night’s 2 goals against OHR). Of the 72 defenders in the NHL that have >100 powerplay minutes, Butcher is 7th in powerplay points per hour. Seventh. In the whole league.
In summary, Butcher generates the most offense of any defender on the team, and is one of the more productive powerplay defenders in the entire league.
These are statistics that are largely dismissed by the analytics community but easily interpretable and appreciated by the mainstream fan and so I decided to include them as part of the explanation for why Butcher may be looked down on. He is the least physical defender on the team by a fairly significant margin with over 3 less hits per hour than every defender other than Mirco Mueller. He also blocks less shots than any defender not named Damon Severson — who has his own detractors with regards to defensive aptitude.
He also has a bit of a reputation for being overly ambitious when moving the play forward. He does give the puck away more than any defender other than Severson and Vatanen — although that may just be a consequence of those 3 being the most frequent puck-handlers in their respective pairings. And so, in a sense, Butcher is the most efficient of those three.
These are some of the stats that people like to point to for defensive performance. Butcher is not physical, does not block shots, and does commit occasional turnovers. This may give the perception that he’s a liability defensively. We’ll see later on if that is, in fact, true.
Advanced (Individual Stats)
These stats are, left to right, game score, shot attempts, expected goals, and penalties. Will Butcher has the highest game score both in rate and cumulative total — influenced by possession, point production, and penalties among other things (write-up here). Butcher is actually not a particularly frequent shooter — he opts to set up more often. Vatanaen, Moore, and Lovejoy (?... again with this) are all more prolific shooters. And it’s not as though he is taking particularly dangerous shots — only Mueller, Vatanen, and Santini have a lower expected goal rates. This is possibly another reason that he is underappreciated — people typically value taking shots more than setting up shots. We use “shots” as a proxy for offensive impact. This is why John Moore has been misconstrued as an “offensive” or, even more egregiously, a “puck-moving” defenceman.
With regards to penalties, he is the most cautious defender on the team with the lowest penalty rate (0.18 penalties per hour). And he draws penalties at the same rate. Of the regulars only Andy Green is also even in penalties — season-long 7th man, Mueller is as well.
Advanced (On-Ice Relative to Teammates)
This is where it all comes out. Butcher is the irrefutable king of on-ice statistics for the Devils. This is to say that, good things are happening at a higher rate when Will Butcher is on the ice than for any other defender. This is not a refutable statement. What we can debate is how much is of that should be credited to him versus his usage.
This chart uses relative-to-teammate statistics. It shows how your linemates improve/suffer from your presence and weights the effect based on the amount of time spent on ice together. For a thorough write-up on its calculation, check out EvolvingWild’s piece on Hockey-Graphs. This chart shows defenders’ effect on their teammates’ shot attempts (columns 1-3), goals (columns 4-6), and expected goals (columns 7-9).
We generate shot attempts, expected goals, and goals at a higher rate when Butcher is on the ice. Defensively, only Mueller and Lovejoy allow see a defensive shot attempt and expected goal results on ice. Those two and Santini have better on ice goal results defensively.
Where critics will make their home is in the context variables. As you can see from the last two columns, Butcher is used heavily in the offensive zone and very seldom in the defensive zone. Not shown here, but also relevant, is that his QoC (quality of competition) is the lowest on the team. These two aspects of analytics are still rather hotly debated. Both are valuable in isolation but not yet shown to be so in the aggregate (see these pieces on zone and competition). Players’ total zone usage — including defensive, neutral, offensive zones and, most commonly, on-the-fly shifts — tend to even out in all but the most extreme cases, as do the quality of skater a given player competes against. Furthermore, the QoT (quality of competition) is more extreme than the QoC and can cancel out those effects. View Butcher’s QoT/QoC graph from HockeyViz as an example of this — slightly low QoT forwards and QoC defenders, extremely low QoT defenders and QoC forwards. Put simply, the low quality of skaters he plays against is unlikely to be MORE impactful than the low quality of skaters he plays with.
To summarize, these context statistics may or may not be impactful. What’s almost definitely true, though, is that they are not be impactful enough to be the sole reason for Butcher’s position atop near all of the on-ice statistical defender rankings.
A3Z Stats (All-Three-Zone Project)
As some of you know, I’ve done some work visualizing Corey Sznajder’s All-Three-Zone Project (support his Patreon here) and you can find the graphs here. Above is the current state of the Devils defense according to these tracking results. Butcher is the most prolific shot contributor and both the most frequent AND efficient zone exiter. He is one of the worse players in zone entries (only clearly better than Green) and he’s slightly below average at entry prevention. Some would point to that last statement as evidence of poor defensive play, I would point to those earlier metrics as not merely mitigating, but completely eclipsing the defensive detriment with offensive boon.
Usage Over Season (using HockeyViz)
Above is a link from Dr. Micah McCurdy’s HockeyViz. As you can see from the top row, his time on ice has barely changed all season, and that level has, more often than not, been the lowest total on the team. As you can see from the bottom row, he’s been given no more defensive zone time. The second row shows that his shot rates have been positive virtually all year, and the third row shows his goal rates had a midseason hiccup which is likely where some of the concern comes from. But should that concern really be enough to warrant keeping a guy the 6th most used defender despite having shot charts that look like this:
I mean come on — those shot maps are Taylor Hall-esque. Keeping that guy off the ice is a crime.
Having seen all of the evidence, let’s now summarize the cases for and against playing Will Butcher more.
Summarizing The Case Against Butcher
- He’s not physical enough to play against tougher competition: He hits the least on the team and is certainly not physical. Whether that means he can’t face top competition is debatable.
- His defensive game needs work, as rookies typically do: He doesn’t get a defensive starts ever, and is still only middle-of-the-pack with regards to expected goals against. He also is below average in defending against zone entries.
- He’s doing well because he’s being sheltered: His 5v5 time on ice was going up rather steadily in the first 30 games — it’s only if you look at the whole season that it appears as a horizontal line. In those 30 games, his expected goal ratio grew the first 20 games or so, and fell the next 15 at which point, his TOI began to decrease (data via OffsideReview). There’s nothing to indicate his QoT/QoC or ZSR has changed or impacted his performance yet. Still, having only played around 40 games a season in college, keeping him fresh and easing him along may be in his best interest.
Summarizing The Case For Butcher
- He is the most gifted offensive defender we have: He generates more shots directly, that is mirrored in his on-ice shot rates, and his point production. This particular bullet is likely the most objectively true statement in either case.
- His ability to push the play forward trumps any perceived defensive liability: His offensive numbers are much more impressive than his defensive numbers are concerning. When factoring both of them, his on-ice relative stats are the best among our defenders. The effect of ZSR and QoC are not well-defined in analytics, but Butcher’s good exit numbers (~top 5% of NHL) would indicate that he actually may be well-suited for more defensive zone responsibility.
- We’re simply better when he is on the ice: As mentioned in #2, he improves everyone around him according to relative stats. Since that can be a bit abstract, here’s another representation in WoWYs via NaturalStatTrick. Of the skaters who have played 100+ minutes with Butcher 5v5, EVERY SINGLE ONE WAS OVER 50 CF% WITH HIM AND UNDER 50% WITHOUT HIM. The universality with which his impact is observable is truly amazing.
Conclusion, and Your Thoughts
I understand why some people are concerned about giving him more time. He’s smaller, less physical, especially around the net, and he’s young. But the league is moving towards skilled young players and Butcher fits the model of one of those guys. At a certain point we need to realize that it can’t be merely coincidence that good things are consistently happening more when he’s on the ice. If we make the postseason, I expect Hynes to treat Butcher the same as always, but if anything I think now is a great time to unleash him. No more preserving him for the long haul, the season ends if you don’t escape the first series and I want the guy that makes everyone around him better out there as much as possible.
What do you guys think? What are the pros and cons? Do you think he should have been played more, but now you’re scared to mess with what’s working? When would you consider playing him more ore less? Would you just wait until next season?
As always, leave your comments below, and thanks for reading!
It turns out I’m not the only one who thinks this! About a half an hour before this was posted, Devils Athletic writer, Corey Masisak, came to a similar conclusion. Subscribers can take a look at that article if you want someone else’s logic on why Butcher should be given more opportunity. There’s some overlap (Micah’s viz, for example), but the pieces were conceived completely independently and coincidentally and so there is a lot of original takes over there too.