Devils in the Details


Are We There Yet: progression, possession, and the New Jersey Devils

The 2017/18 season has been a time of radical change for the New Jersey Devils organization. The emergence of a new wave of rookies, including Nico Hischier (2017, 1st overall, QMJHL Halifax), Will Butcher (signed 2017, free agent, NCAA Denver), and Jesper Bratt (2016, 162nd overall, Allsvenskan AIK), is one of the more obvious examples of radical change.

Yesterday, the New Jersey Devils traded forward Adam Henrique - a fan favorite and a player capable of playing across the lineup, and the scorer of one of my favorite goals of all time. His departure, while difficult, is further evidence of transformation the organization is currently experiencing under the direction of General Manager Ray Shero. Only two players remain from the 2012 playoff run - Andy Greene and Travis Zajac. 

Perhaps the most radical change of them all is the sudden appearance of praise for Devils hockey. Across the league, markets large and small are talking about the Devils without using the word "Trap", except in the past-tense and with a sense of incredulity at the stark difference between this team and the Devils of old. Sporting News ran an excellent piece on Shero and the Devils, citing a belief that the Devils have turned, and are on track for a successful future. Beyond that, the article pointed to the current season as an ongoing success. 

Without a hint of irony, Adam Wylde of the Steve Dangle Podcast described the Devils in their most recent episode as "Fast and Attacking", hitting two of the three pillars of the current organization's philosophy:

  • Fast

  • Attacking

  • Supportive

Like most others, I was skeptical of this new approach when GM Ray Shero introduced coach John Hynes to the media and described this new philosophy. I was excited, mind you, for the change itself, should it ever actually take place. We had been treated to the stylings of Michael Ryder, Damien Brunner, and 1PP specialist Jordin Tootoo. At that point in time the Devils had been trying to fill holes with veteran free agents, Wiley AHL all-stars (AKA Bobby Farnham), and Tuomo Ruutu. The future was relatively bleak beyond Schneider, Larsson, and a select few core pieces. 

Looking from then to now, it's incredible to see what were once seen as likely buzzwords have started to materialize on the ice. The team is fast, the team is clearly aggressive and attacking, and sometimes even supportive.  This is the most fun I have had watching Devils hockey since Henrique, Brodeur, Kovalchuk, and Salvador brought the team to the finals over five years ago. 

There is only one problem - I'm not sure we're quite there yet.

On Being "Actually Good"

Watching the games, there is no denying the talent on the Devils roster. Instead, the tiny voice in the back of my head can't help but look at some basic metrics and say "I'm not so sure about this". 

Whether you are a stats person or not, there is a clear and logical correlation between possessing the puck, and being successful in the long term. You need to have the puck to score. Our methods of tracking possession are imperfect and sometimes over-simple, but they are good at drawing out basic narratives - the story of what your team is accomplishing on the ice. 

In recent memory, the Devils have been one of the worst possession teams in the National Hockey League, measured by Corsi (and all of it's sub-metrics), also called SAT. Corsi is a very simple metric that is made up of shot attempts. If your team directs more shot attempts at the opponent than your opponent directs towards you, you have a positive Corsi differential (Corsi For minus Corsi Against) or Corsi For Percentage (percentage of total shot attempts during the game).

If you only measure unblocked shot attempts, you are measuring Fenwick.

In brief, the Devils were, and still aren't, a good possession team as measured by these simple metrics. While the Devils are winning, their plain old CF% is ranked 30th, at 46.29%. This means that only 46.29% of shot attempts are headed towards the opponent's goaltender. While being 3.71% below an even 50 may not seem like much, the spread of Corsi is small, usually ending up within 2-3 percentage points of 50. And since we're talking about long-term success, when over the course of the 2016/17 season the Devils gave up 3490 shot attempts, percentages and scale matter. Plus there's the fact that nobody wants to be 30th. 

On Episode 14 of the Devils in the Details Podcast, I mentioned my concern about the Devils' possession this year, and some very basic insight into why it might not be as bad as it seems. 

There's the obvious answer, which is that we're winning. We have Cory Schnieder and Taylor Hall at the top of their games, and a collection of excellent young rookies. The cliche, but perhaps appropriate, answer to this is that the Colorado Avalanche were winning several years ago when they rode an awful CF% and sky-high PDO to the playoffs, only to fall back to earth for the next two seasons. What I want isn't the playoffs once. I want consistent, stable, sustained success. 

A more interesting explanation involves differentiating between shot attempts, because as may be obvious, not all shots (or attempts) are created equal. Corsi, as a basic metric, doesn't differentiate between a shot from the slot and a shot from center ice. Both are recorded as a 1 in the ledger. 

However, hockey analytics have evolved to the point where we can differentiate between low-danger, medium-danger, and high-danger shots and/or shot attempts. When I say that the Devils are sporting a 46.29 CF%, these variations are not represented. This number represents CF% across all three types. 

When we look closer and examine only high-danger shot-attempts, which are usually identified by location, we see something different. The 2017/18 New Jersey Devils have a 52.41 HDCF%, good enough to place 9th in the league. The team's high-danger CF% is over 6% higher than their general CF%. This is an astounding difference. 

Because I want to believe that the Devils's early success, while likely the result of some over-achieving, is deserved, my instinct is to attribute this difference of 6.12 percentage points to "The System". It might be the case that the style of play the Devils are using is good at limiting shots and attempts to low-danger areas. It seems obvious to me that any good system would aim to limit high-danger shots, but maybe the Devils are just better at it than other teams. 

The Devils have continued their tradition of having excellent goaltending under the Shero regime. The team's save percentage so far this season has been an excellent 92.89%. In recent memory, a "league-average" goalie has been roughly identified as 91.50%, plus or minus a few points. However, the team struggles more than most in high-danger situations. Their high-danger save percentage (HDSv%) is 86.52. This is a large and statistically significant drop. 

Here's a simple chart outlining the differences in both CF% and Sv% across the two categories discussed above:

Cory Schneider is not above this issue himself. His save percentage, when divided up from low to high-danger, looks like the following:

  • LDSv% - 96.02

  • MDSv% - 96.58

  • HDSv% - 81.40

Luckily, the Devils are good at limiting high-danger shot attempts. High-danger shots, or successful attempts, represent only about 25% of his workload. However, Schnieder's HDSv% does put him at around 30th in the league. 

Are the Devils "Actually Good"?

First off, I am not nearly smart enough to answer that question, and I think it's honestly a silly question to begin with. We can identify trends and averages, and we can employ the eye-test, but Devils fans should know that it only takes a bit of chaos to take a team farther than the numbers suggest

What I think these particular numbers suggest is that the Devils are employing a system that plays to their strengths, which for the first time in a while are speed and offense. I think a spread of over 6% between CF% and HDCF% tells a story of a team that is able to keep things to the outside, even if that means that the opposing team gets more shot attempts throughout the game, every game, for all 82 games of the season. As Ian suggested on the podcast, maybe that means that when the opposition does get a high-danger attempt, it's somehow worse than plain old high-danger - some sort of really-high-danger (RHDCF%?). 

Perhaps this is all an illusion, and my creeping fears are correct - the wheels are going to fall off at any moment and we'll find ourselves in what I call the Buffalo/Arizona/Vancouver-style purgatory - sucking, but never bad or lucky enough to get the highest draft picks. This could be the case, but I don't think so. 

I'm starting to believe. Maybe the Devils are onto something, and know that they need to do everything they can to keep shots to the outside, and somehow it's all working. Maybe they know that in the offensive zone, their system of employing elite offensive talents like Nico Hischier, Taylor Hall, and Brian Gibbons (?) is enough to overcome a general imbalance of shot-attempts overall. 

The season is just over halfway finished, and the Devils might actually be good. 

Duncan Field