clock menu more-arrow no yes

Filed under:

Devils Skaters’ Performance in “All Three Zone” Data

New, comments

Using zone entry tracking data, I profile all of the Devils skaters and comment on how well they transition the puck forward. Plus, pretty pictures!

NHL: Arizona Coyotes at New Jersey Devils Ed Mulholland-USA TODAY Sports

In this article, I will be showing you guys the results of a new viz I’ve been working on that I published last week. The viz concerns results of Corey Sznajder’s All-Three-Zone tracking project. He tracks varying aspects of zone entries, exits, entry defense, and passing. In order to fully understand the implications of the viz and the terms used therein, I decided to define a few of the stats and give a crash course on what we know about them.

All Three Zone Primer

1) Zone Entries

Zone Entries are likely the most studied of the stats that you will see in this piece. Many including JenLC13 have found that controlled entries (carries as opposed to dumps) are where it’s at, and AAtJ alum, Ryan Stimson, found that successful passes after entries entries heighten the predictive value even more. Charlie O’Connor concurred with Ryan that total entries per 60 was not repeatable. He lobbied for replacing it with Weighted Entries per 60, a stat which includes both uncontrolled and controlled entries and controlled entries. In most cases, ratio was as important or more important than rate — which is to say controlled entries per entry attempt are meaningful as well.

For the purposes of my viz, you will see 2 terms:

PossEntries/60 = How often per 60 minutes does a player enter the zone with possession? Calculated as 60*(Entry Passes + Carry-ins)/TOI

PossEntry% = In what percent of attempts, does the player retain possession? Calculated as (Entry Passes + Carry-ins)/(Entry Passes + Carry-ins + Dumps + Fails)

2) Other Transition Stats (Zone Exits and Zone Entry Defense)

Other transition stats are not studied quite as much, but there is still some things we know. I use zone entry defense and zone exits in my vizzes so I’ll explain them a bit here. In both cases Original Six Analytics found them to be helpful when predicting CF%. In particular, controlled exits/60 and controlled entries against/60 were most powerful. In fact, Viz Wiz, Sean Tierney, and PDOCast host, Dimirti Filipovic, collaborated in visualizing the combination of exits and entry defense into a net Net Neutral Zone Score and examine some of it’s behaviors. For the purposes of my viz these are the terms used.

PossEntries/60 Allowed= How often per 60 minutes does a player enter allow a zone entry with possession? Calculated as 60*(Entry Passes Allowed+ Carry-ins Allowed)/TOI

PossEntry% Allowed = In what percent of entry attempts in which a player is targeted, does opposing player retain possession? Calculated as (Entry Passes Allowed + Carry-ins Allowed)/(Entry Passes Allowed + Carry-ins Allowed + Dumps Forced+ Breakups)

Breakups/60= How often per 60 minutes does a player break up a zone entry when they’re targeted? Calculated as 60*Breakups/TOI

PossExits/60 = How often per 60 minutes does a player exit the zone with possession? Calculated as 60*(Exit Passes + Carry-outs)/TOI

PossExit% = In what percent of exit attempts, does the player retain possession? Calculated as (Exit Passes+Carries) / (Exit Passes+Carries+Dumps+Clears+Icings+Fails).

3) Shot Contributions

These are tracked in the image of Ryan Stimson’s original Passing Project, premiered here at AAtJ. There is a lot of work on the different aspects of the project and you can browse over at meta-hockey if you want more info on anything discussed here. But suffice to say, what happens before the shot is currently not officially recorded, and also hugely important. Ryan has shown that incorporating types and quantities of passes preceding a shot, we can greatly improve out modeling of expected points. Examples of such pieces can be found here, here, and here. However, in the interest of presenting only easily interpretable information, I restricted the info to shots and shot assists.

Shots=Shot attempts. Calculated as (Missed Shots + Blocked Shots + Shots on Goal)

Shot Assists=Passes directly or indirectly to a player who got a shot attempt. Calculated as Primary Shot Assists + Secondary Shot Assists + Tertiary Shot Assists.

Shot Contributions = The total number of shots a player contributed to creating. Calculated as Shot Assists + Shots

Devils Players in All Three Zones

Alright, now that that silly boring “defining terms” section is out of the way. Let’s get to our Devils. I’ve been making visualizations, similar to the HERO charts by Dom Galamini, of the zone stats listed above using data tracked by Corey Sznajder (support him on Patreon, and follow him on twitter). You can view the viz used for this article here. What you will see in each of these visuals is the skaters’ percentile position in each stat within their position (forwards or defenders. For instance, when you see Miles Wood get an 81 in “Shot/60,” that means that he has higher shots per 60 than an estimated 81% of forwards, putting him in the top fifth of the league in the stat. For reference, 289 defenders and 548 forwards have been recorded.


Something that I’ll say at the onset here of analyzing this pictures is that, contrary to the “All Three Zones” moniker, they are definitely weighted towards players who generate offense and/or move the puck in transition. Therefore, guys that profile as defensive zone specialists -- limiting dangerous attempts, but not gaining possession — are not going to show well. Also, this doesn’t represent all games. The sample sizes are listed below their names and so a lot of these numbers could change with more data. Bearing that in mind, there are still a few things we can pull here.

#1 Our first line is stacked. Taylor Hall and Nico Hischier are elite at moving the puck into and within the offensive zone. The both fall in the top 20 in average percentile. To be clear “highest average percentile” doesn’t hold any real statistical significance — just describes what the “bluest” graphs would be. Bratt isn’t elite, but he doesn’t present as having any real liability. Johansson, Gibbons, and Coleman are all not generating much moving into and within the offensive zone, but they are all pretty good at exiting the defensive zone.

#2 Wood is the only other dynamic player. I would put Kyle Palmieri in the “will hopefully improve with larger sample” bin of players. Zajac is around average at % categories, but low in per 60 categories, indicating he doesn’t take part in moving the puck forward much, but when he does, he’s average at doing so. Noesen just shoots, Hayes is average at everything. Zacha is very efficient at entering the zone, but doesn’t do it much and is average at everything else.

#3 Stafford should probably not be waived. There’s obviously more to this story than just these charts, but Stafford is comfortably the 5th in line on our team in shot contributions. The first line and Wood are the top 4. Stafford is basically tied with Jesper Bratt. They are in the 66th and 67th percentiles respectively, meaning the contribute to shots at higher rates than about two-thirds of NHL skaters. Next on our team is Stefan Noesen who is comfortably below the league median at the 31st percentile. Stafford is also perfectly acceptable in transition, and better than much of what we put out on a nightly basis in moving the puck forward. His other peripherals are not great, so this might require further investigation, but it’s hard not to notice him here.

#4 Brian Boyle is among the least dynamic players in the NHL. That’s not to say that he is the worst player in the NHL, definitely far from it. It’s just this aspect of his game that doesn’t seem to have much hope of improving at this stage of his career. He’s poor in every domain, and as seen in the tweet above, has the lowest average percentile score in the NHL. He does not move the puck forward at all. His faceoff success, physicality, and net-mouth presence, make him a perfect example for a player that contributes in other ways. But there’s just no way of framing these results to make this a “good news” situation for him.


Oddly, the numbers for our defenders seem slightly less embarrassing. You probably wouldn’t know it to if you only saw our “top 4,” though. For the purposes of this article I’m calling Greenetini the top pairing because up until very recently, they were routinely being given the toughest assignments (recently they’ve split and Santini was sent to the AHL); and I’ll be calling Vatamoore the 2nd pairing.

#1 Top Pairing struggles offensively. Andy Greene and Steve Santini enter the zone infrequently and inefficiently, and don’t generate offense once there either. Greene is good at exiting the zone, Santini excels in preventing possession entries, but neither can flip that around and force the puck up the ice in the other direction.

#2 Second Pairing isn’t much better in those areas. John Moore is a volume shooter and he’s exceedingly mediocre in everything else. He’s doesn’t break up many entries at all, allows a lot of attempts, but he also doesn’t allow a disproportionate among of controlled entries per attempt. Sami Vatanen’s possession entries/60 allowed is much better, and he also is competant at entering the zone.

#3 Butcher and Severson excel at moving the puck forward. Will Butcher is one of the best defenders in the league at exiting the zone with the puck. He ranks 20th in the NHL in possession exits per 60. Damon Severson is good at a lot of things. Unsurprisingly, he and Butcher are the best passers, and he is also the most prolific Devils D-man in zone entries, More surprisingly, though, Severson and Santini are the toughest to gain the zone against. This is among the reasons I’ve called Severson our all-around best defender.

#4 Small sample weirdness. Mueller shows as one of the best defenders in the league at exiting, and shows well in entering as well. He may be good at it, but being this good is likely a product of small sample quirks. Also Lovejoy showing well in offensive stats and poorly in defensive stats has small sample written all over it. Worth noting here that Vatanen has small sample as well, and even smaller when you consider the first 2 tracked games were with Anaheim.

Concluding Thoughts

The lopsidedness of our forwards in these stats is what’s most striking to me. Hall and Nico are all-NHL in transition. Wood and Bratt are also quite good. Everyone else is in the red in almost everything. Just in case you’re curious if that’s normal, it’s not. I can’t imagine being that dependent on a 4 guys to move the puck forward is a sustainable system. With regards to the defense, I continue to think that the coaching staff overvalues minimizing dangerous attempts within the defensive zone, and undervalues preventing time in, and moving the puck out of, the defensive zone. The latter skills prevent attempts against and cause attempts for, whereas the former skill does the reverse.

Again, be wary of the sample sizes in those images, and don’t over-extrapolate their implications. These microstats are relatively new and really are best suited to contextualize other info — they can often help bridge the gap between analytics and the “eye test.” That being said... Have fun.

Thanks for reading and leave your comments below!