Actually Above Expectations: Shooting & Scoring by the 2018-19 New Jersey Devils

Those who have followed the New Jersey Devils understand that the team is far from a group of defensive stalwarts. The goaltending has not been good. The defense has suffered big time from critical errors. They have even found unconventional ways to make games harder for themselves such as goaltender poking pucks to opposing players and scoring goals on their own net. With this known, it is important as ever for the Devils to keep scoring goals if they want to stay in games and even try and win a few of them. To that end, let us look at how they are shooting now - which is actually a little bit above expectations.

Sean Tierney, @ChartingHockey on Twitter, has been doing a lot of great work with his public tableaus and charts based on the NHL’s play by play logs and other statistical functions. With shot location data in the meta of those play by play files, Tierney and other resources (e.g. Natural Stat Trick, Corsica) can identify how many scoring chances a team has in a game or how many times the team gave up a scoring chance with a player on the ice. A model for goal scoring based on shot location has been developed in past seasons called the Expected Goals model and it has been a staple in analytics since then. What Tierney does is plot out all of the shots on a tableau by location while listing the player’s stats: how many goals they did score, what their expected goals number was, their average shot distance, and so forth. The data Tierney uses is for all situations, so it includes power plays, extra-skater, and overtime situations. It’s also up to date; I looked it up yesterday and it does have the San Jose game. It’s a great tool and there’s a lot to learn from playing around with it.

For the purposes of this post, I used it to look up every Devil and how they have shot the puck this season. I split it up between defensemen and forwards because, as you may expect, the defensemen have typically shot the puck further away and from a specific area than the forwards. To highlight who is excelling or falling behind the expected goal count, those skaters who are more than a goal below their expected goal count are highlighted in red and those who are above it by a goal are highlighted in green. The expected goal model is just that: a model. It is not a perfect system and it is not meant to taken as a guarantee that so-and-so should have X goals but they have X+Y goals. It is meant to provide a baseline of how often a player should have scored based on where he took the shot and what type of shot he took. I will use such a baseline for all of the Devils skaters. Even those in Binghamton or not on the team anymore; they’re in grey and in italics.

As an aside, those who want to take this to the “next level” and are willing to put in a lot of work should consider the third dimension. That’s not being counted by the scorers at NHL goals, so you’d have to put in a lot of work in yourself to track it somehow. But even a generic “low, medium, high” for every shot attempt could yield a massive breakthrough for the expected goals model and how we understand shooting on a statistical level. Don’t say I don’t have any ideas - even if they are remarkably impractical or tedious to take on for now.

The Shot Map Data for Devils Defensemen as of Dec. 11, 2018

Sean Tierney Shot Map Tableau

The Devils defensemen have not scored a lot of goals. A grand total of 8 have been made and half of them are by Damon Severson. Since this does include power play situations, that Severson has been set up in the right circle has given him closer shots than just hanging back at the points. Further, Severson (and on a few occasions, Egor Yakovlev) have been willing to jump up on offense to contribute. That helps too. All the same, Severson is clearly the best among the blueline when it comes to scoring and putting shots on target.

Curiously, as a group, the defensemen have 358 shots on net. The team has taken 1,315 shots in total this season. Defensemen make up just under 30% of all of the team’s shooting. But because a lot of those shots are from distance and often they’re going into traffic or they’re shooting through a clear shooting lane that the goalie sees through, a lot of them fail. So the Devils defenders have contributed a little over 9% of all goals. If there’s a conclusion to take away from this, then it’s that the overall offense may be better served by the defenders passing the puck to someone closer to the net rather than fire 50+ footers at the net and hope and pray they get in somehow. Alternatively, the defenseman can also try to get in closer to help themselves out.

That said, only one defenseman can really say they have been unlucky with goals and that’s Sami Vatanen. The model suggests that even with all of his shots from the right point, he should have 3 goals. He has one. Granted, Vatanen’s shots (and the other defensemen) may have created goals through deflections and rebounds. But Vatanen’s shots have not created many rebounds and his on-target percentage (shots on net over attempts) is second only to Severson. Given how he has taken some risks by carrying the puck more on defense and willing to try to make a move with it himself, Vatanen’s shooting in general would be better served to move himself closer for shots. What he has been doing is either going to garner him some breaks or have him continue scoring below expectations.

The forwards, well, they’re a more interesting group. And more important since they have scored the vast majority of New Jersey’s goals this season.

The Shot Map Data for Devils Forwards as of Dec. 12, 2018

Sean Tierney Shot Map Tableau

Let’s start with the good.

• As you may expect, Kyle Palmieri and his team-leading 16 goals lead the way in this category. Palmieri has been firing them above expectations over the whole season. It is true that it has come in bunches. Check out his Game Log at NHL.com. He scored seven in his first four games and two more later in October; he had just three in November; and he scored four in back-to-back games prior to the loss at San Jose. That’s the thing about most scorers. Sure, the best of the best do it with more consistently; but it is more common that the goals come here and there as opposed to every night. All the same, Palmieri’s stick has been hot over the whole season because when he gets one, he tends to get another not too far in the future. Fortunately for the Devils, Palmieri’s volume of shots is high and his on-target percentage is not too bad. Keep going, #21.
• This partially explains why Travis Zajac has not been reviled by the fans so far this season. Zajac has produced above the expected goal model and arguably expectations in general. Given how his wingers tend to change and how he was moved up in the lineup as needed earlier this season, Zajac’s utilization has been shifted around. Still, he has been quite good at getting relatively close to the net, which is generally a good thing. Only Nico Hischier has been closer in terms of average shot distance. Zajac also leads in shots off rebounds, which has no doubt contributed to the scoring. May it continue!
• I was surprised to see Brian Boyle also beating the expected goal model for his shots. It is not by as much as Palmieri or Zajac, but for the role Boyle plays, it is worthy of being highlighted. Whatever you can get from a third/fourth line forward can help make a difference for an offense and Boyle has done that. I recognize that he did score three of his seven goals in one wonderful night against Pittsburgh. But as I pointed out with Palmieri, goals are not usually scored consistently by the majority of players. By the model, Boyle would not have had that so it is worth appreciating.
• Coming in just ahead of my one-goal-over-the-model’s-number rule is Jesper Bratt. Bratt has been a bright spot through some dismal weeks. Since his return to the rink, Bratt has been a mainstay of the top-six and has generally been an effective forward. Bratt may not be shooting the puck a lot and his shooting percentage is rather low. However, based on where he’s shooting it, Bratt has done better than what has been expected. I would like to see Bratt take more initiative to shoot from better locations. To get more goals, I think he’ll need to be a little more judicious than firing from any sort of angle from the wings. Otherwise, it has not been bad from the young winger.

Most of the remaining forwards have scored about as many as the model suggests. Of note among them:

• Pavel Zacha could also stand to be more judicious about the kinds of shots he does take. While he had some horrible luck early on, the model suggesting he should only have about three goals means he could do a bit more.
• On the flipside, Blake Coleman has been shooting the puck well but not so well that he’s blowing the model out of the water with a crazy shooting percentage. Some of those shorthanded breakaways and odd-man rushes may contribute to his expected goal number so more volume may be in order from the Texan.
• Some fans have wanted to see Hischier and Marcus Johansson shoot more, well, they may have a point. Hischier really does get close to the net for his shots, which speaks to his bravery and willingness to get in the “dirty” areas on offense. He just needs to do it more, really. Johansson has been more accurate than I expected. 76 shots in 26 games is a higher shooting rate than he has had in previous seasons, so his volume should be seen as a plus with respect to his past. I suppose he could just use some better puck-luck. Ditto for Hischier to a degree, too.

Now onto the bad, or the players in red. I’m not so worried about Stefan Noesen or Taylor Hall, albeit for different reasons. Miles Wood will get his own section.

• Noesen is a depth winger and, like with Boyle, whatever you can get from them is a bonus. He’s also shooting below 3% this season, which is really unfortunate. However, the expected goals model suggests he should have only one or two more goals than he does; he’s averaging just over two shots per game; and he’s not particularly accurate when it comes to taking a shooting attempt. But given the limited action Noesen normally sees at even strength, he does not get a regular shift on either side of special teams, and he has been healthy scratched, I do not think Noesen is hindering the offense.
• Hall, on the other hand, deserves more leeway. His shooting percentage is not bad as it is above 11%. Hall leads the team in shots and by a fair margin, so it is not a case that he is lacking in volume. He still has 8 goals when the model suggests he could have 10, which is not a huge differential. Moreover, Hall still has a point-per-game average of over one. Hall has been involved in many goals by other Devils, particularly Palmieri. I’m confident Hall will get a few more breaks to boost that goal total in the future. Hall gets top minutes, he plays well in 5-on-5, he receives power play time, and he contributes a lot more than just scoring goals.

Then there’s Miles Wood. Oh, Miles Wood.

The Miles Wood Issue or Let’s Look at Some Actual Maps Instead of the Numbers

Based on the expected goal model, Wood should have about six goals. He has two. He has the lowest differential between actual goals and expected goals at -4.181. Wood is shooting at a woeful 2.7%. For comparison’s sake, Wood shot at 11.2% last season. There is reason to believe he is the unluckiest Devil on the team.

This is where Tierney’s tableau shines. The data here is pulled from a map of shots. Here’s Wood’s shot map for this season so far. I highlighted where the scoring chance area is normally recorded.

Sean Tierney Shot Map Tableau

Look at the spray of his recorded shots. Wood has taken a lot of them from the left wing and behind the dot. Plenty are above the circle, so Wood was effectively firing shots from the left point. He has taken plenty of angled shots that he has not scored at all on. On the offwing, there’s a smattering of similar shots.

Wood does plenty of shots in the “house” so to speak. It is not like he has not taken some attempts that can be classified as scoring chances. But look at the color. The light blue represent snap shots, the dark blue are backhanders, and the purple are wrist shots. I’m seeing more wrist shots outside of the scoring chance area. In that area, there are more tips, backhanders, and snap shots. While the scorer may be mis-classifying shot types, Wood really should be firing a more reliable shot like a wrister instead of snapping one off or attempting a harder-to-control shot like a backhanded shot.

For contrast, here’s the shot map for Palmieri:

Sean Tierney Shot Map Tableau

While Palmieri has taken his fair share of long-distance shots, there is a lot more action and more variety in shot-type in the scoring chance area. There are several purple marks for wristers. There are not so many backhanders, which makes sense as Palmieri’s best asset are his forehanded shots. What’s more is that even outside of the scoring area, Palmieri has taken several attempts from inside the circles. A lot of those red marks - slap shots - are in his usual spot on power play situations. While technically outside of the area, those are not bad shots to attempt. Notice that there are few attempts beyond the outside hashmarks on either side. He has not and does not take many shots like this; he’s usually funneled toward the net. Which is what he should do as a shooter and does do.

I know Wood is not the same player as Palmieri. He is not utilized like Palmieri. He does not play with the likes of Hall, Hischier, Zacha, or Johansson like Palmieri has this season. He does not get as much power play time as Palmieri. However, Palmieri’s shot map reflects a forward who knows where his shots should be taken. Wood, in contrast, is more all over the place and I think that contributes to Wood’s lack of goals this season.

Wood’s map suggests to me that Wood’s decision making on when, where, and how to shoot the puck could be better. Wood has speed and strength on his side, so it does not make sense at a macro level that he settles for shots from the wing in not-so-ideal locations off the rush or just because he has some space. I do believe Wood has been unlucky at shooting the puck to a point. After all, even with his shot map, the model suggests he should have about four more goals than he does. But the map itself shows that Wood can stand to make some improvements for the benefit of himself and the team. Wood needs to take more shots in better locations, take more wristers in those locations, and look to make a pass instead of firing a long shot he is not likely to score on.

One Last Sobering Point

Tierney does not just make maps of the shot type and shot location for each player. He does not just have the data pulled from those maps. He also has team data for expected and actual goals. The sad thing for the New Jersey Devils is that they collectively have 88 goals. The expected goals model has the team at 86.71. That’s right, the team is a little above expectations for scoring for this season so far. Even with Wood, Hall, Noesen, and Vatanen scoring more than goal less than the expected model. Palmieri, Zajac, Boyle, and (to a lesser extent) Bratt combined drove the total differential to be ahead of the model.

This supports what I wrote at the very beginning of this post. The Devils’ goaltending has mostly been in a range from not good to utterly abysmal with good performances occurring few and far between games over the last two months. The defense has been enough to quell the opposition. They have been regularly guilty of errors that only make their goaltender’s job that much harder. When a team is forced to play defense for stretches and/or concedes a bunch of goals, that also keeps the offense from having the ability and/or will to push forward and create the offensive opportunities they need to stay in games and/or get results out of those games. The good news is that the Devils are not so unlucky or unfortunate with respect to scoring based on where the team is shooting it. They should be seen as being a little bit above expectations, in fact. The bad news is that it is not enough. The Devils need to make gains in where they are weak and should try to keep improving on offense to generate more opportunities to score to make up for those weaknesses.

Based on this post, there are some ideas just by looking at Tierney’s shot maps. The Devils defensemen could stand to defer from taking long distance shots and try to make more plays that create closer shots. Most of the forwards could stand to keep doing what they’re doing. Some are hot and should definitely do that. Some, not so much, but they’re not doing something terribly wrong. The one who stands the most to gain is Wood and by trying to shoot in more central locations and with more wrist shots, he could help himself as opposed to his current state of just firing all kinds of non-slapshots where ever he is. He’s no Kyle Palmieri, but he could take an idea or two from him about shooting in more advantageous spots.

Your Take

Go check out Sean Tierney’s work. This is his shot map tableau and it is fun to dive in and see a graphical representation and a data summary of a player’s shots. Tierney has other public tableaus available that are worth exploring. So please do that, if nothing else.

In the meantime, what do you make of all of this? Are you surprised that the team has scored better than what the expected goal model suggests? Do you think Palmieri will keep beating the model? What about Zajac, Boyle, or Bratt? What do you think Wood needs to do to improve his scoring? Are you also convinced from this that the defensemen should take fewer shots from the point? What else have you learned from this post? Please leave your answers and other thoughts in the comments. Thank you for reading.