Not Enough Danger: Another Look at the 2018-19 New Jersey Devils Offense

All shots are not created equal. The expected goals model accounts for this by determining value for all shots, mostly driven by its location and type. Last week, I wrote about the expected goals of the 2018-19 New Jersey Devils players. Scott Martino had the following comment:

And since the Devils had a low projection to begin with, perhaps it stands to reason that goalies gifted them more goals than they should have this season. Would have to see the save %’s by danger area against the Devils to make that determination.

The expected goals model does not really go into specific danger areas. Its determination . However, what Scott is referring to is another way to define shots. Shots can be binned into low, medium, and high danger based on its location. Natural Stat Trick has a great and straight forward explanation in its glossary that goes into what makes a shot a high-danger as opposed to medium or low danger.

There’s a larger argument between binning and non-binning, but in my view, there is value in these definitions as they refer to very specific types of shots. Basically, if you’re getting looks at the crease or rebounds that bounce into the slot, then discretely counting that from other shots has value that you would not get from expected goals. Anyway, as I wrote about the Devils’ xG last week, this will be about danger. I had a post back in February about what the Devils allowed with respect to danger; this will be about what the Devils generated. Or the relative lack thereof.

The Overall Team Stats or Get Ready for a Lot of Red in this Post

Hidden away in the Show/Hide Columns button, the team stats at Natural Stat Trick have full counts of shot attempts (a.k.a. Corsi or CF in this case), shots (SF), and goals (GF) by low, medium, and high danger. Since their definition of danger is straight-forward, I will be using them as a source for all stats in this post.

Unfortunately, I have yet to find a source that breaks down attempts, shots, and goals by player. Natural Stat Trick has individual high-danger scoring chances and individual xG, but I cannot tell you who on the Devils scored the most goals from high-danger shots or who on the Devils generated the most low-danger shot attempts. We will have to work with team stats instead. If you know of a site that has these kinds of stats for players, then please let me know and there may be another follow-up post.

Anyway, I pulled 5-on-5 and all situations team stats from Natural Stat Trick. The 5-on-5 situation is the most common in hockey and all situations are included for further perspective. I’ve included how the Devils ranked in each stat, highlighting those in the top ten in green and those in the bottom eleven in red. Get ready for a lot of red. Here are the total team stats regardless of danger.

Data from Natural Stat Trick

I used Sean Tierney’s expected goals model last week, so Natural Stat Trick has a different result for the team’s xG. Still, it is one of the smaller ones in the NHL. The Devils may have outscored the model in 5-on-5 and all situations, but it was not by much and it was still not a lot of goals. The xG value points to how much their offensive work would have yielded in theory. The Devils’ shooting percentage was not terrible but not all that good either. Having both a sub-7.7% in 5-on-5 play and a sub-9.3% all situation play for shooting percentage suggests that the Devils could have used a little more better puck luck in general.

The bigger point is that the Devils also lacked in volume. The Devils were among the worst teams in the NHL last season in terms of generating shot attempts. Finishing next-to-last in total shot attempts suggests multiple issues. Off the top of my head, I question whether zone entries put the Devils in positions to shoot, whether the Devils were good enough at moving the puck forward, how the Devils were coached to attack on offense, and whether they could maintain possession and drive play in general. I think it is all of that and more. The Devils managed to put approximately 57.10% of those attempts on net - the best proportion in the league. But even that led to a total shot volume and shot rate that ranks below most NHL teams. You need to generate possibilities for shots to get shots. Efficient as the Devils were at putting their attempts on target, they needed more attempts.

Now, this is the team summary. How does danger play a role? Let’s start with the level with the most surface area and least threatening.

Low Danger or I Guess This Went Well But...

Data from Natural Stat Trick

Hey, its the color green! Get used to it here. Out of the three danger levels, the Devils’ shooting percentage for low-danger shots ranked highly in the league. They scored many goals from low-danger chances. This went well - for a given definition of well. Per Scott’s original question, this suggests that the Devils have had a few gifts in their favor.

As you can see from the total count, racking up 28 goals in 5-on-5 situations out of low danger shots is still not very many goals. It is just 19.1% of all 5-on-5 goals scored by the Devils. While the shooting percentage ranked well, it is still just 3.28% out of all shots. It is generally good to be one of the better teams in the league in a stat. However, this one does not contain a lot of value. Even here, you can see the issues of volume. The Devils ended up around the league median for low-danger shots in both 5-on-5 and all situational play. The Devils still ranked low in generating low-danger attempts.

Given that these shots are often long ones from the point, distant ones from the sideboards, and sharp angles by the goal line, it is not really a surprise that a vast majority of them do not become goals. What surprised me is that the Devils attempted as many as they did. I checked the other 30 teams and they all had way more low danger shots and low danger attempts than the other two danger levels. Why?

I have some thoughts. First, it may be by definition. Natural Stat Trick’s definition of low danger also includes shots from medium danger areas that were blocked, which may also raise these numbers. A team collapsing on defense or a defender filling in a shooting lane is common; I can see blocks juicing these numbers. Second, it may be by design of the game. Hockey is a game where players have to make split-second decisions with the puck and this does result in less-than-desirable plays like just firing away from 50+ feet away or just flinging a puck towards the net by the boards on the outsides of the faceoff circles. Non-ideal decisions are constantly made, which adds further chaos and variation in this glorious game of skill, speed, and intensity. At least the Devils are far from alone in having a significant proportion (around 45%) of their total attempts and shots being classified as low-danger.

Now what about the next level; the shots from the the scoring chance “house” that are not at the crease or off rebounds or rush plays? Get ready for more cells highlighted in a light red.

Medium Danger or Getting Denied by the Goalies and the Defense

Data from Natural Stat Trick

For the medium danger shots, few of them became goals. In 5-on-5 play, goalies stuffed the Devils not-very-high number of medium danger shots. Just 37 goals in a bin where the league median in this category has 45 does not seem like much, but the ranking is clear that it is something. Most teams were able to get more production from these more favorable locations. Most teams were able to generate more attempts from these more favorable locations. The Devils being as accurate as they were with attempts helped them finish in a more respectable ranking for medium danger shots. The rate, well, it was decent for 5-on-5 and definitely not for all situations.

Similar to low danger shots and attempts, most of a team’s goals are not coming from these kinds of shots. Since they require being closer and they are in areas where teams are more protective, there are not as many attempts or shots. However, being in better positions helps yield more goals. Even a high low-danger scoring team like New Jersey scored more goals off of medium danger shots than low danger ones. There is some value to improving in this area.

But its high danger where the Devils need to do more. Alex wrote yesterday that, compared to high danger shots and attempts against the Devils, the Devils seem OK. But that is more of a function of how well the Devils have prevented those kinds of chances.

High Danger or A Reason Why Hall Clearly Said to NJ Needs More Talent

Data from Natural Stat Trick

By Natural Stat Trick’s definition, the high danger shots are ones at the crease or shots in medium or high danger areas off rebounds and rush plays. Even with such a definition and a generally lower fraction of high danger shots out of all shots, look at the goals. Most teams get a significant amount of their production off close shots, shots off rebounds, and shots off the rush. Teams look to counter-attack and/or put an emphasis on transition offense to get those rush plays. Players crash the net hoping for loose pucks to bang in; the old cliche of “going to the net” is justified in part by these results. Even for the Devils, the majority of their goals scored come from high-danger chances.

Even so, the Devils were quite bad at this relative to most of the NHL. A shooting percentage around 16% is not that good and given the types of shots high-danger represents, I wonder if there is a finishing issue than otherwise. Opposition goalies did relatively quite well off of high-danger shots. Yet again, that brings me back to the issue of volume. The Devils may have been great at limiting the other teams from high danger attempts and shots, but the Devils were not at all great at creating them. Even with a higher shot-efficiency (around 75%), the Devils did not register a lot of high-danger shots.

The whole table is red for a reason; the Devils really struggled when it came to generating high danger attempts and finishing high danger shots. While I believe that the coaches may be able to address this to a degree, this is where the lack of offensive talent really comes into play. It is not just in finishing more high danger shots, but being able to make the plays to make them possible. This can be making a great breakout pass that catches the opposition unaware. This can be stopping an opposing player on defense and immediately starting a counter attack. This can be having the speed to stretch a play out enough to get that killer pass across the slot during an odd man rush. My point is that it is not just generating the high danger chances for yourself, but generating for others too.

When Taylor Hall said the team needs more talent, these stats indirectly back him up. As much as coaching has a hand in a lot of this, these are the significant plays where having someone who has special talent can make a big difference. (There are at least two players in this year’s draft that has that and that is partially why they’re at the top of the class.) Let me put it another way. Natural Stat Trick does have individual high danger scoring chances by player. The Devils had one (1) Devil put up over 100 in all situations last season: Nico Hischier with 108. Just four other Devils had over 50 iHDCF. For perspective’s sake, Carolina had four players with over 100 iHDCF and eleven with over 50 iHDCF. There is a bit more to it than just get better guys and watch the team have more tantalizing opportunities to score, but it is needed for the Devils to be better in this part of offensive stats and in offensive stats in general.

General Conclusions & Your Take

Let us go back to Scott Martino’s question. Did the Devils get gifted more goals than they should have? If we assume that goals from low danger shots are gifts, then the answer is yes. The Devils were one of the more high scoring teams from low danger chances last season. Do I agree with the assumption? Not completely. A goalie guessing wrong on a save attempt, getting frozen on a shot, or a straight-up fluke can happen regardless of where the shot was taken. Unless I re-watching every goal scored by the Devils and counting any bad goals allowed by the opposition, I cannot say how far off it is or not. So I can understand those who want to make that assumption that LDGFs are gifts for the Devils.

Regardless, they represented a small percentage of all goals scored. NHL teams get most of their goals from high danger shots and their medium danger shots tend to yield more goals than low danger shots. This was true for the Devils; however, they were one of the league’s lowest scoring teams in those two levels of danger. Worse, they were one of the league’s worst at generating attempts of any level of danger. Even with an efficient shot-to-attempt ratio, the relative lack of attempts held back the number of shots on net - which hindered the number of shots that would go in the net.

For the Devils to be better next season and in seasons beyond, the Devils have to figure out ways to generate more offense and add to the team’s talent level to execute the plays that lead to more desirable offensive opportunities. Taylor Hall, who will be a part of the solution when he is healthy, was right to demand more talent. The first overall pick of the 2019 NHL Draft will do this. And I anticipate that general manager Ray Shero will do more in the offseason to do just that. Hopefully we can see some gains made and sustained by this time next year.

Now I turn to you, the reader. What do you make of the Devils and danger from last season? It was not good. What would you do to make it better other than draft a top tier offensive forward in the 2019 NHL Draft? Please leave your answers and other thoughts about the kinds of attempts and shots last season’s team took (and did not take). Thanks to Scott Martino for the question. Thanks to Natural Stat Trick for having the team stats that go into all three levels of danger as well as the clear definition for each. Thank you for reading.