Visualizing Corsi and the New Jersey Devils

The other day Tyler Dellow posted an entry to his mc79hockey blog which was entitled Visualizing Corsi and the Jets. His inspiration was a good article written by Tim Bonnar at Arctic Ice Hockey. The article attributed the Jets woes this season to a lack of depth, specifically the bottom six forwards. mc79 created a chart visualizing the 5v5 corsi% for any combination of forward and defenseman who had played a minimum of 10 games and 60:00 minutes on the ice together.

I have since seen the chart replicated several times. First, by Cam Charron for the Vancouver Canucks. In the comments section of Cam's post, I noticed links to the same chart for the Blue Jackets and Rangers. Although, I am not sure if these last two examples filtered by at least 60 minutes of TOI as a minimum benchmark, because both charts are completely filled in, which would be unlikely. I decided to put together the same chart for the Devils because I thought it would be fun to see how it shakes out.

The premise is as follows, stated originally by mc79:

If we assume that the coach uses his best players the most, we would expect that the Corsi% for when the most heavily used defenceman and the most heavily used forward are on the ice to be the highest on the team, barring something really unusual – a defenceman who plays exclusively in the defensive zone or something. … If my theory’s right, things should get worse as we move from more frequently used to less frequently used players.

I decided to do two charts for the Devils with the exact same data, except colored differently in each case. I think most of us are familiar with Corsi by this point, but in case not, Corsi for is the total number of shot attempts generated by the Devils (shots + missed shots + blocked shots) when a certain player is on the ice. Corsi against is the same thing, but for the opposition. The number in each box is the Corsi for % of the two players when on the ice together. Corsi for % =Corsi for / (Corsi for + corsi against). The first chart is colored red for above 55%, orange for above 50%, and blue for below 50%. Please keep in mind none of these numbers are weighted to take into account quality of competition or anything like that. It is just a direct look at 5v5 corsi % for individual F / D player combinations. The forwards and defensemen are both listed in decreasing order of TOI / 60 at 5v5.

All of the data in the charts is taken from Hockey Analysis.

What we see for the Devils here is a bit different than any of the other teams I've looked at.

So much red in that chart. How does this Devils team not have more wins again? Interesting to see what a boss Andy Greene is in general, or what a dominant situation it is when you have Greene, Mark Fayne, or Eric Gelinas backing the Devils top line and top 6 forwards. Basically the Andy Greene column shakes out exactly as you would expect. The play is tilted most in the Devils favor when he is on the ice with the top forwards, and the CF% decreases as you work your way down through the lineup, although not by much until you get down to the fourth liners. The Devils top forwards pretty much come out ahead no matter which defensemen they are paired with, although to a lesser extent when playing with Marek Zidlicky. You can also somewhat see visual evidence here of the struggles of Adam Henrique so far this season at 5v5. There is more blue in his chart then any other Devils forward listed.

Also, I think this chart shows how heavily the Devils have really been leaning on their top 3 lines in terms of ice time, at least in 5v5 play. First of all, there are only 11 forwards on this list. I couldn't even find a twelfth forward who played at least 60 minutes on the ice with even one Devils defender. That's because typical twelfth forward Cam Janssen plays about 2 minutes a games, and a few others have shuffled in and out of the lineup due to injury. Although, I will say, I don't think depth to be the Devils problem when analyzing this chart. The top 9 forwards pretty much all come out ahead when paired with the top 4 defensemen.  And, it's not like it's an absolute tire fire when you get below that threshold - in fact everything looks pretty good, with one or two exceptions.

Since the Devils are such a strong possession team, I decided to show a chart that is colored based on the CF% relative to the Devils team percentage. I thought this would provide a visual of if a certain combination is helping or hurting the team relative to the overall standard.  Perhaps with a team like the Devils, merely coming out 50% or greater is not so great when the Devils team corsi for % is 53.8.   In this chart, I would expect the color coding of the boxes to look very close to the original hypothesis.  The players the Devils are putting out on the ice the most should be helping the team the most, relative to the others.

If a combination has a number higher than 53.8%, I colored the box red. If it's lower, I colored the box blue. It would stand to reason that the people who see the ice the most should be helping the team the most.

I wouldn't say that this chart shook out exactly as I expected either.  No one looks good with Peter Harrold in this chart. Interesting that Zidlicky is a little bit below the team average when playing with the Devils top line, since Jagr, Zajac and Zubrus are pretty bossly. Mark Fayne fares better when playing with the top forwards, as one would expect. Marek Zidlicky is an interesting case, as he actually seems to fare a little better when he plays with worse linemates, as Ryan deduced in his post from Monday illustrating a new stat, CC%, or corsi contribution %.

Through his tracking of passing data and shot attempts generated, Ryan is able to deduce just how much of a direct impact each player has on driving play forward - is the player creating the chances, or just a passenger while others do the work? Unlike my charts here, which are really just showing already available data in a more visual way, Ryan's contributions are pretty groundbreaking. If you haven't taken the time to read his post on CC%, please take the time to do so, it is well worth it.

Here is an excerpt from Ryan's post showing how important Zidlicky is to the Devils offense, regardless of the fact that his Corsi % may not be as high with the top players.  It doesn't mean, in Zid's case, that he is not helping the team.  He certainly is.  It just seems he is better suited to be, perhaps, away from the toughest competition the other team has to offer.  Hence, he looks better when not paired with the Devils top forwards.  Please note that these numbers are through the 30 game mark:

Now, Zid’s iCF was 60 for the first 30 games. His SAG was 55. So, we can say that Zidlicky was directly responsible for 115 shot attempts (60 of his own shot attempts and 55 he created by completing passes to others that resulted in a shot attempt). What can that tell us about how important he is when he’s on the ice? Of all the 399 shot attempts the Devils attempted when Zid was on the ice, he was responsible for 15% via his shot, and 13.8% via his passing, for a combined 28.8 CC%. When Zid’s on the ice, nearly 29% of the team’s shot attempts go through him in some way. That is 2nd among defensemen to only Eric Gelinas..

Interesting thing about Eric Gelinas, he is driving the play much more by bombing away, since he is responsible for 85 total shot attempts, 73 of which are his own. Here you have two different Devils players, both which are driving play forward directly more than any other on the team. We can see here that nobody on the Devils blue line creates more with his passing than Marek Zidlicky.

No surprises on the Andy Greene front. He also fares better when playing with the top forwards. It would seem that the Devils top forwards paired with Andy Greene and Mark Fayne should do well in most matchups against whatever any other team can throw at them. Eric Gelinas is once again looking good with every single forward he played at least 60 mins of 5v5 with. Anton Volchenkov is another player who seems to fare better when not playing behind the Devils top forwards. Perhaps this is also because it allows him to face weaker competition in these scenarios, while the Devils top line usually faces the best the other team has to offer.

So overall, the chart did go somewhat as expected, although not entirely. Play does appear to tilt heavily in the Devils favor when Andy Greene or Mark Fayne are on the ice with just about any Devils top forwards. Eric Gelinas creates a ton just by bombing away more than any other defenceman. Marek Zidlicky's case was a bit more unusual. Combine that and the struggles of Adam Henrique, and I don't think this chart shakes out quite as neatly as some of the ones I've seen for the other teams. It is clear that the Devils are deeper on defense and in forward depth than the Winnipeg Jets. It is also clear that Jack Johnson is really really awful. The Canucks defense is deep across the board and their top 6 forwards are very strong. However, it would appear that the Devils may have a bit more forward depth than Vancouver (albeit a bit less top end talent). Anyway, this would be something pretty interesting to look at later in the season or after the season is complete, as 60:00 of TOI is still a really small population size in the grand scheme of things. It would also be interesting to create a chart that was weighted for TOI, perhaps with different sized bubbles such as used on Hockey Abstract. But, that will all have to potentially wait for another day.

Now, I'd like to know, what, if anything you take away from these charts. Do you agree that this shows how dominant the Devils top end forwards can be? What do you make of Zidlicky faring better when not playing with the top line? Do you see the Devils as potentially having more forward depth than most of the other teams looked at so far, at least in terms of controlling possession and the run of play? Anything else you see here as far as trends or things that we can take away from these visuals? Please take a moment to leave your take in the comments section, and thanks for reading.