Greene: The Last Defender
Last year was, predictably, a trying one for the Devils’ blueline. Outside of Andy Greene, the Devils haven’t really had a top pairing defender for a LONG time. For team that once boasted a top 4 of Scott Niedermayer, Scott Stevens, Brian Rafalski, and Paul Martin — all of whom were, at some point, individually capable of being a #1 defender — it’s been extremely tough sledding the last decade or so. Ever since the last of those 4, Martin, left the team, it’s been Andy Greene and ...
Andy held down the fort admirably for a stretch. In fact, according to WAR metrics, he was an elite NHL #1 defender. On Manny Perry’s WAR model (writeup here) Greene was the most valuable defender in the league from 2012-2014 (via Corsica), and on the Younggren twins’ model (RITSAC slides here) he was 4th in the NHL behind only Marc-Eduoard Vlasic, Mark Giordano, and Alek Pietrangelo (via Evolving-Hockey). This viz by Alan Wells, shows how great he was at his peak in both metrics.
He topped out at basically the 99th percentile. Now, the models disagree as to the precise anatomy of his subsequent decline, but, in both, it appears that his last 3 years have been those of a slightly above average defender (~60th percentile) whereas the previous 3 had been those of an elite defender (~80th percentile). EvolvingWild have said that it takes 3 years of data to feel confident in a WAR assessment, so the rough agreement on this trend is concerning.
Int particular, last year, other metrics took a slide. John wrote a piece on Greene’s decline here. I slightly disagreed, but, too, conceded that he was no longer a top pairing defender. It was on him this year to prove that he could still be a top 4 guy, despite his detractors and the looming Father Time.
Severson: A Prospect Awakens
Severson has a different story. He was a member of a group of defensive prospects that was, at the time, one of the most encouraging in the league. It included the likes of Alex Urbom, Jon Merrill, Eric Gelinas, and Reece Scarlett. As you probably know, almost all of those didn’t pan out. They played a combined 34 games last season — all by Merrill. As the lone standing player from that once proud prospect pool, Severson rode early success to a 6-year contract. He was off to another strong start last year, resulting in an endorsement from your’s truly that he was the best defender on the team. From that article to the end of the season, though? He was anything but (sorry!).
From that publish date on, Severson was a negative in CF%Rel, a heavily SCF%Rel, and the 5th worst defender in the NHL in HDCF%Rel (via NaturalStatTrick). Was this something that was a troubling hiccup in his development, or was he never that good? On WAR metrics, Severson has never been as good as he was his rookie season, and even then he was roughly the 30th percentile of defenders. I pointed out the conflicting narratives on Severson’s analytics this past offseason — if you believe that his performance last year was anomalous, it’s likely because you believe John Moore dragged him down.
Severson’s Old Partner, John Moore: The Phantom Menace
After splitting the first half of the year pretty evenly with Greene, Moore, and Butcher, around the same time I wrote that piece, Moore became his consistent partner. If you analyze Moore’s impact on Severson’s shot metrics over the past two years, it’s clear that he hurt in almost every category except one ...
Interestingly, Andy Greene had the opposite impact. He helped in the general shot metrics, but hurt in goal totals.
It’s my best guess, that the coaching staff was biased towards these goal results, and that led them to overlook the more reliable statistics which present a reality in which Greene and Severson was a pairing that actively helped both players. Look below at their WOWYs:
As you can see, John Moore was completely dependent on Severson to inflate his metrics across the board. All 4 stats were increased when he was with Severson. In comparison, all of Severson’s shot metrics fell (as mentioned before, goals rose, IMO causing bias). The Severson-Moore pairing consistently inflated Moore’s numbers and deflated Severson’s. Contrast that with the Greene pairing, which consistently increased both of their shot results. This phenomenon persists despite the fact that Severson-Greene actually drew tougher matchups with lower zone-starts than Severson-Moore.
The goal of pairings should be to raise the sum total of the teams performance. A pairing that hurts two players is obviously terrible — but a pairing that helps both (Severson-Greene) is just as obviously preferable to on that only helps one (Severson-Moore).
2018-19: The Return of Greenerson
So, coming into this season, with Moore now gone to the Bruins, Severson’s 2nd most common (and MUCH more effective pairing) was potentially set to resume. Mirco Mueller politely acquiesced by being a serviceable partner for Green’s ex, Sami Vatanen — this scenario likely doesn’t occur without his contributions. This left Greene, Severson, Butcher, and Lovejoy as the remaining 4 defenders. With Butcher-Lovejoy as one of the top 10 most efficient defensive pairings last year, the pairing decisions were obvious.
Ben Lovejoy and Will Butcher picked up right where they left off — the Devils have scored 4 goals an allowed none with them on while also continuing to register strong shot metrics. Meanwhile Andy Greene and Damon Severson have been one of the most effective defensive pairings in the NHL. According to Corsica’s pairing tool, the have the 7th highest xGF%rel (expected goal ratio relative to team) while scoring 5 goals and allowing 2 and being a +3 in penalty differential. This pairing has been excellent for the Devils.
So, the next question is “Why?”
Why should a young D who plays like a misplaced forward, and an aging, has-been, stay-at-home d-man combine to be one of the most effective pairings in the league? There are a few possible explanations.
1) They Complement Each Other’s Skills
Each of the two players have glaring holes in their game. Andy Greene is terrible at driving play in transition — he’s a poor passer, shooter, and skater. He seldom drives exits/entries and often cedes at an alarming rate as well.
Damon Severson is not only strong, but the Devils best defender in overall transition metrics. Surprisingly, he’s even strong in the defensive ones. Greene will be able to play a sound defensive game, but then dish the puck over to Severson, who is a prolific exit/entry defender.
Meanwhile, Severson’s glaring deficiency was overall net-mouth presence. He cleared rebounds apathetically, refused to use his frame to box forwards out, and often got caught out of position in critical moments. Among Devils defenders with 300+ minutes over the last 2 seasons, the only player with worse SCA60rel and HDCA60Rel (scoring chances and high danger chances allowed per hour relative to team) was ... John Moore (seriously why did they think that was a good idea?).
Conversely, despite having some of the most difficult assignments in the league, Andy Green was excellent at preventing high danger chances. Out of the 291 defenders with at least 200 minutes, Greene was 6th in the NHL in the WAR component “Quality Against.” This indicates that after controlling for the total amount of shots against, the average danger of them was among the lowest in the NHL. Severson? He was 289th in that metric ... 3rd worst in the NHL.
Each of them are not only lacking, but among the worst defenders in the league at those respective skills. Conveniently those skills happen to be their partners’ forte.
2) Andy Greene Is Better Than We Thought — Easier Deployment Unleashed Him
(*Note: For anyone unfamiliar, QoC=Quality of Competition, QoT=Quality of Teammates, ZS%=Zone Start Ratio)
Above is a tableau graph of the defenders (1000+ minutes) with the most difficult assignments over the last 3 years. The QoT-QoC bar is comprised of the z-scores of QoT and QoC metrics for TOI and xGF% and the ZS% bar is made of the z-scores for OZS% and DZS%.
What was that whole chunk of gibberish I just spouted. Doesn’t matter. The long and short of it is this: Over the last 3 years, Andy Greene has had the most difficult deployment of any defender in the entire NHL. Given the tread on the tires, and his aging, last year finally seemed like a bit too much for him to handle.
Fast forward to this year — Andy Greene is being given the most favorable zone deployment of any defender (possibly any skater) on the Devils team. The QoC/QoT also no longer leads the team, though he’s still clearly 3rd behind only Vatanen/Mueller and not that dissimilar from last year. He’s still getting decently difficult matchups (tough opponents), but getting them in DEMONSTRABLY easier situations (many more offensive zone starts).
It’s possible that what these WAR numbers were uncovering was that, despite the more conventional analytics (Corsi-type metrics) deriding his performance, he was doing fine given how difficult his situation was. And now that his responsibilities have gone from most difficult in the NHL to actually slightly favorable, Greene is proving he never lost the step we claimed he did.
In this scenario, Severson likely has Greene to thank for the success of this pairing.
3) Severson Is As Good As We Hoped, and Thrives With Anyone Not Named “John Moore”
According to Evolving-Hockey, Severson’s been hovering at just below 0 in RelTMxGF% stats his whole career — so it wasn’t too surprising that that happened again last year. However, his RelT CF% had decreased every one of the 4 years he’d played. Of the 20 Devils defender seasons (500+ 5v5 minutes) from 2014-2017, Severson produces 3 of the top 4 RelT CF +/- but has been falling back to earth. Is it a coincidence that his first season was hist best one, and also happens to be the only season in which John Moore wasn’t a Devil? Perhaps. Or Perhaps not.
It seems unlikely that Severson, one of the most egregious allowers of high-danger chances over his career, has suddenly gotten notably better at that particular skill. However it does seem possible that his transition skills (which are sometimes more predictive than existing shot metrics), combined with favorable usage, and a partner that doesn’t drag him down so much, that he is driving possession so much that the chances against simply don’t have the opportunity to materialize.
In this scenario, Severson is likely the main cause for the success of the pairing.
I think it’s a little of all three things actually, but mainly the first one. Damon Severson has the highest RelT CF% with +7.21 by a country mile (Butcher is +2.58), and Greene’s +12.3 RelT xGF% also leads by a lot (Seves is #2 with +7.73). In other words, it seems like both are still excelling primarily in their area of expertise, but the other has picked up the slack in the problem areas. Severson now has the lowest RelT xGA/60 on the team which is EXTREMELY uncharacteristic of him, and Greene has the highest RelT xGF/60 which is equally absurd. Put into normal-people English, Andy Greene’s offensive numbers and Damon Severson’s defensive numbers are both basically the best they’ve ever been thus far.
It’s clear to me that Greene’s time as the minute-munching, situation-proof, top-pairing guy was over. The lightening of the load helped him. It’s also pretty clear that John Moore was a leech on the underside of Damon Severson and his leaving had a positive impact. However, it also seems to be that 6-28 just work.
And so now that I’m closing in on 1800 words about this topic, I look forward to the lineup for the Nashville game tomorrow having already broken up the pairing because John Hynes doesn’t like it when I’m happy.
What have you thought about 6-28? What do you think is driving their success? How about the rest of the pairings? If something needs to be toyed with during the season, what might that be?
Leave your comments below, and, as always, thanks for reading!