To give a very brief intro to what’s going on in this piece: it’s become something of an annual tradition here at AAtJ for someone to go over whether or not the Devils were “lucky” in the previous season.
I looked at it last year, and Alex had looked the year before. The very basics are that Rob Vollman (hockey analytics pioneer and current LA Kings employee) had a system of assessing luck that included features that I’m going to paraphrase here:
- Puck Luck — How many more goals did the team score/allow than expected? Formerly assessed using PDO, updated here to use xG vs Goals differential.
- Injuries — How many injuries did the team have? Formerly assessed by CHIP (cap hit of injured players), updated here to GAR Lost.
- Special Teams — PP and PK situations are less repeatably than 5v5 and so ST (special teams) performance is a measure of luck to some degree. It is assessed here using PP% and PK%
- One-Goal games — “Close” games (especially OT/SO games) can often go either way due to variation and so poor record in those situations is indicative of bad luck. Here I’ll look at 1-goal game record alone, and versus their record in other games.
The Devils were pretty clearly not lucky when it comes to the “percentage” game. Keith Kinkaid was one of the worst goalies in the league last year, and Cory Schneider was terrible early as well. The injuries to forwards also likely drastically reduced the overall shooting talent which may have contributed to the fairly severe offensive underachievement as well. Combining these two components, even adjusting for the relative quality of the shooters and goalies, the Devils had bad puck luck.
The Devils lost a ton of value in injuries — the 7th most in the NHL according to this, and I’m not even sure that does it justice. Unlike many of the other teams, the Devils faced the vast majority of those injuries all to one position (forward) simultaneously. That meant that not only did the lose value in the injury, they also lost value because the replacements were worse than your typical replacements. I looked into this when counting up the GAR values of the major added and subtracted assets. As such, I’m fairly confident in saying the Devils had bad injury luck.
The Devils were 21st in PP% at 17.7% (League Average: 19.8%) and 4th in the NHL in PK% at 84.2% (League Average: 80.2%). This story is really all about the penalty kill. The Devils have a rich history of elite penalty-killing, and that continues today. John talked about the effectiveness of the forwards earlier this offseason, and I focused on the Lovejoy-Greene pairing which — the greatest PK pairing of this era — in my farewell to “Rev.”
However, unlike last season, when Taylor Hall dragged the PP to competence on the back of one of the best single-player PP performances in the league (a feature that likely buoyed his Hart campaign considerably) his injury — and many others, as noted above — really tanked the PP effectiveness. Now, it’s admittedly difficult to ice a competant PP unit when your most skilled healthy forward is Blake Coleman, but nonetheless, this is a feature that we would expect to improve next season.
In total, the Devils finished 9th in Special Teams differential. The Devils PK has been good for a long time, but with Greene aging, and Lovejoy gone, it’s fair to wonder if we can remain league-best. That said, it would be surprising of a healthy and juiced up powerplay wasn’t above average in the league given the potential personnel. I think all-in-all, you have to say the Devils had roughly neutral special teams luck last season.
The Devils had a 37.5 winning percentage in 1-goal games. That the 4th lowest in the league, but it their winning percentage in other games was only 38%. They were 3-4 in the Shootout and 4-6 in OT so only slightly below average in those two categories despite being an awful team. Most of the time So, while their results were poor, they weren’t entirely dissimilar from their overall performance, and so I’m going to say the Devils had roughly neutral luck in 1-goal games.
The Devils were bad. Don’t misunderstand the findings of this piece as a counterargument. However, using these metrics, it seems that “neutral 2018, unlucky 2019” is a more viable explanation than “lucky 2018, neutral 2019” which means that even without the additions, we’ve expect at least a little regression to the mean through normal variation.
What do you guys think? Did you feel last year was a uniquely unlucky season? Were we just bac? Do you expect luck to improve this year? Thanks, as always, for reading, and leave your thoughts in the comments below.