Yesterday, I made a With Or Without You (WOWY) analysis of Zach Parise. The original WOWY analysis was done by Tyler Dellow of mc79hockey, where he utilized statistical scripts from Vic Ferrari's Time on Ice resource to determine the Corsi events for a player with and without his teammates. The analysis focuses solely on even strength, 5-on-5 situations with a goaltender in each net from the regular season. What we can conclude from a WOWY analysis is who has the largest Corsi impact on a player.
From what I did in my WOWY analysis for Zach Parise, I concluded that Travis Zajac has been the most ideal forward for Parise to play with (and vice versa) in terms of Corsi, that the defense in general benefited greatly with Parise but suffered without him, and from Parise's Corsi perspective, Jamie Langenbrunner and Dainius Zubrus would be suitable linemates on the opposite wing. Most of all, Zach Parise had a positive impact on nearly all Devils skaters who have had 100 or more Corsi events with Parise in 2009-10. Essentially, the WOWY analysis shows in another way how valuable Parise was last season to the New Jersey Devils.
This stat would also help players that have struggled find themselves better when put together with the right players. If this isn’t too time consuming, I would love to see Rolston’s, perhaps Langenbrunner’s, and even Kovy’s WOWY stats too see exactly who they mesh with the most. (I only say Kovy’s because I know some don’t think his style belongs here and the chemistry was shattered a bit).
Zach Parise finished 2009-10 with 2,357 events (1274 for, 1083 against, 54.1%) in 81 games. Ilya Kovalchuk joined the team on February 5, played only 27 games, and so ended up with 792 events (398 for, 343 against, 52.9%). It's not the same proportion as Parise, but it's not terribly far off and that's pretty impressive given that both players played the same position.
Let's see how Kovalchuk impacted his teammates' Corsi in those 27 games. From there, we can determine who suffered the most when they were away from Kovalchuk and who's absence saw Kovalchuk's Corsi suffer the most in his abbreviated time as a New Jersey Devil. We can see how Corsi can suggest who he fit the best with as Devil, and should he stay, who should be on the ice the most when he's there.
Again, please view this post in Wide before you continue after the jump. Please select it in the gray box next to the headline of this post to make it Wide.
Before I jump into Kovalchuk's numbers, here's a side note about Corsi.
A Side Note: Corsi Isn't Everything
There are some base criticisms with using Corsi alone. Here's a quick refresher of what Corsi is: A Corsi event takes place when a shot on net is made, a shot misses the net, or a shot is blocked. When you're on the ice when a Corsi event happens for your team, it's a positive Corsi event; it's a negative Corsi event if you're on the ice when the other team has an event.
While it suggests territorial advantage and offense, it isn't uncommon for teams to have puck possession in the other team's zone (which is good) but aren't able to attempt a shot for whatever reason - but do benefit from just spending time in the other team's end. Such possession isn't captured by Corsi.
It does not take into account the quality of opposition or teammates. It does not take into account what the score is, the talent of the player, the role of the player whether the player is tired, or anything like that. It counts shooting attempts with the assumption that good possession will lead to shooting attempts. A Corsi percentage is positive Corsi over of the total Corsi, so it is desirable to have a high percentage; especially for an offensive forward. A lower percentage for a defenseman or a defensive forward may come up but it doesn't necessarily mean they were bad.
It's a good stat but it shouldn't be the only stat that one looks at, nor should it be taken without some kind of context such as ice time, games played, points scored, quality of competition and/or teammates (Behind the Net is an excellent resource for that), etc. With a WOWY analysis, we can at least take other factors into account so it's not totally devoid of context.
For Ilya Kovalchuk, since he only played 27 regular season games with the Devils, he's not going to have as many events as Zach Parise and given the initial issue of fitting in with a new team, his percentage may not be as high as a player who has been with the team all season.
Ilya Kovalchuk's WOWY with the 2009-10 New Jersey Devils
Amazingly, when I pulled the numbers, Jay Pandolfo didn't show up as players who played with Kovalchuk even though he did play in some games toward the end of the season. Odd that they didn't show up but Andrew Peters did. So they are the only ones not included in this analysis. I don't think it matters since when they did play, they were on the fourth line anyway - Kovalchuk wasn't going to have much interaction with them at all.
Here's the full team chart with Kovalchuk's WOWY. It's so big, so it's not embedded here - please click on the link to view the chart.
Similar to what I did with Zach Parise, I've grayed out all of the names of Devils that had fewer than 100 Corsi events with Ilya Kovalchuk. Among the grayed out group, you'll find that Zach Parise was the only top 6 forward to not have interacted with Kovalchuk all that much in even strength, 5-on-5 play. Sure, they saw plenty of each other on the power play; but they were kept separate namely because they play the same position. It made little sense for Jacques Lemaire to put one of them on the off-wing just to have them together; either way, it represents one combination the coaching staff didn't really bother with too much.
Forwards and Zajac - 100+ Corsi Events Together
In fact, only 5 forwards have had 100 or more Corsi events with Kovalchuk in his 27 games as a Devil. Yes, for all of the line combinations and switching, Lemaire seemingly kept groups of forwards together.
The negative WOWY in the first yellow column (teammate without Kovalchuk) indicates that the teammates were better off in terms of Corsi percentage away from Ilya Kovalchuk. You'll notice that Patrik Elias, Travis Zajac, and Dainius Zubrus all have negative percentages. Before jumping to conclusions, I believe some context is in order for those three players. Yes, their Corsi suffered but let's remember that if Elias or Zajac wasn't centering Kovalchuk's line, he was centering Zach Parise's line, a player with whom their Corsi percentages improved greatly.
With respect to Elias and Zajac, who were the Devils' top two centers in the final third of the season, it appears that Kovalchuk's own Corsi suffered more away from Elias than it was from Zajac. Moreover, Elias and Kovalchuk together yielded a better Corsi percentage for Elias than Zajac's Corsi percentage when Zajac played with Kovalchuk. Not that either way bad, but in terms of who did better with Kovalchuk, Elias has an edge based by Corsi percentage together. Yes, he benefits greatly away - much more than Zajac - from Kovalchuk but that's because the other option was playing with Zach Parise, who really driven his Corsi percentage up.
Therefore, based on Corsi percentage, there was some justification for Elias to remain as Kovalchuk's center as well as for Zajac to be Parise's center. Not that Corsi suffered so much when Zajac centered Kovalchuk (I think .529 Corsi% together isn't bad), but the overall benefit of the team was greater with that set up.
It can be said that Jamie Langenbrunner was far and away a more effective right wing for Kovalchuk's line than Dainius Zubrus based on Corsi. Zubrus' Corsi was the worst among all five of these forwards with Kovalchuk, and both he and Kovalchuk's Corsi percentage improved drastically when they were apart - moreso for Kovalchuk. Quickly going back to Zubrus, Corsi percentage justified the two players playing on different lines.
I think these numbers match up with what one notices in their games. Zubrus thrives upon puck possession down low and cycling, whereas Langenbrunner is more versatile in terms of offensive role. Therefore, it didn't surprise me to see that Langenbrunner's and Kovalchuk's Corsi was reduced when they were apart.
It did surprise me to see that Kovalchuk's Corsi suffered more when he was away from Langenbrunner than the other way around. For a winger who's production went dry by the end of the season (1 goal, 0 assists in final 6 games) and had some kind of beef, it didn't bother Kovalchuk's Corsi too much. I can only wonder how much more fruitful the pairing would have been if Langenbrunner didn't start slumping among his other issues.
Lastly, Lemaire did experiment with Brian Rolston on Kovalchuk's right wing for a few games and it was great for Rolston's Corsi percentage. Both players suffered away from each other and in big percentages, but that's largely due in part of how large the Corsi percentage was when they were together. Given the relatively small population size and I'm hesitant to say that Rolston really was a good right winger for Kovalchuk. Maybe the experiment should have continued, especially in light of Langenbrunner's slump at the end of the season. With more Corsi events together, it'd be a clearer argument.
Still, what I really gather from this as well as the WOWY I did for Parise is that Lemaire's top two lines of Parise-Zajac-Zubrus and Kovalchuk-Elias-Langenbrunner were somewhat justified from a Corsi percentage standpoint.
Defensemen and Kovalchuk - 100+ Corsi Events Together
Like Parise, since the defensemen are on the ice for longer shifts and are changed at different times than forwards, seeing the top 6 compile 100 or more Corsi events with Kovalchuk on the ice should be of no surprise. Unlike Parise, the Corsi percentage of only some defensemen were hurt when playing without Kovalchuk. Bryce Salvador and Andy Greene saw huge drops in their Corsi percentage when not playing with Kovalchuk; and Kovalchuk's own Corsi percentages suffer away from them. It suggests that this pairing - both kept together before and after Paul Martin's return - really meshed well with Kovalchuk.
The same can't be said for Mike Mottau and Colin White, where all parties saw improvements in their Corsi percentages away from each other. Perhaps it's due to that neither Mottau and White are offensive defensemen. At least with the Greene-Salvador pairing, Greene has decent offensive skills from the point and can pinch and contribute some decent shots. Salvador, well, he can just throw pucks in if kept at the point.
Incidentally, Martin Skoula's Corsi percentage was negatively impacted without Kovalchuk, but Kovalchuk's Corsi only suffered by a smaller magnitude. This is odd given that his defensive partner for most of his tenure was Paul Martin, a player who actually saw a slight improvement to their Corsi without Kovalchuk. What makes it more odd is that Kovalchuk's own Corsi suffered more away from Martin. Both players didn't have has many Corsi events with Kovalchuk as the other four defensemen, but I'm really curious as to why there were two different reactions to their Corsi with or without Kovalchuk (and vice versa). Especially considering that the other four defensement saw similar effects to their Corsi with their usual partners.
Main Conclusions
I don't think the WOWY analysis definitively shows whether or not Kovalchuk was a good fit in New Jersey. It did show who he did work well with in terms of Corsi percentage change: Jamie Langenbrunner, Brian Rolston, Andy Greene, Bryce Salvador, and Martin Skoula.
Patrik Elias and Travis Zajac performed well together with Kovalchuk, but their own Corsi improved without Kovalchuk. Since they were both centers at the time, only one of them could benefit from having Parise on his wing. The only forward with whom Corsi suggested was a bad fit was Dainius Zubrus. Kovalchuk didn't mesh well with Mike Mottau and Colin White either; but again, that's largely because if they aren't on the ice for Kovalchuk's line, they may very well be there for Parise's line. By the time playoffs started, Corsi percentage does justify Lemaire's selection of Parise-Zajac-Zubrus and Kovalchuk-Elias-Langenbrunner when taking this WOWY and Parise's WOWY into account.
Going back to the context of role, Kovalchuk and Parise are both scoring left wings. They were going to get prime minutes and always be in the top two lines. Therefore, the number of forwards they interacted with on a Corsi level were limited even though they largely played with the most common defensemen lineup at the over the same time period. As much as Devils fans may complain about line changes, it's not as if it was a revolving door at forward or as if Kovalchuk had any say on the defensemen behind him at even strength.
What the WOWY analysis shows that, in terms of Corsi, Kovalchuk did not have a big impact on every single Devil who has had 100 or more Corsi events with him. Together with these players, only a single teammate yielded a sub-50% Corsi percentage with the two together (Zubrus); yet 6 teammates were better off in terms of Corsi without Kovalchuk and only 4 of those 6 teammates did Kovalchuk see a benefit to his own Corsi away from them (White, Mottau, Zajac, Zubrus).
So, no, Kovalchuk didn't have a Parise-like impact on the team in terms of Corsi. We can guess as to why that was during the 2009-10 season, but ultimately it suggests that the coaching staff needed time to figure this out. This isn't unreasonable for bringing in an important player into the lineup. Fortunately, by the end of the season, it appears that the coaches did the best they could and their top 6 made sense accordingly. If Kovalchuk does resign, it'll be interesting to see whether or not the Corsi percentage with some of these teammates would improve with improved familiarity.
Still, assuming Corsi percentage as a rule of thumb, I would suggest to the new head coach that if retained, Kovalchuk's center should be Elias over Zajac (again, since Zajac and Parise suffer much more apart in terms of Corsi) and Langenbrunner or Rolston should be considered for the right wing position.
What else did you gather from this WOWY analysis? If you have any questions or need any clarifications, please feel free to ask. If you think I've made a mistake, let me know and I'll try and fix it. Again, thanks to Tyler Dellow of mc79hockey for helping me out with the Parise study, user njdNYG'cuse for suggestion a WOWY analysis for Kovalchuk, Vic Ferrari for the scripts at Time on Ice, and you for reading.