Coaching Corsi

[jc17]

Whether I've looked at team corsi numbers or individual numbers I've noticed many of the top players are from the same teams ie: BOS, LA, Chicago and NJ

So my question is: do you think corsi is more of a coaching thing than a player thing? There's no way that it just so happens that all the best corsi players are on certain teams.

Click to read more...

 

[charlie1]

The extent to which the coach is responsible vs. the talent on the team itself is difficult to parse out. You would need to perform a statistical analysis which shows, for the same level of talent, certain coaches produce higher CF%. I know of no such study, although it is conceivable that coaching systems have an effect.

With regard to LA or Chicago-- those teams are stacked with players who have always had a high CF%. In some cases the players were chosen because of their CF%.

 

[Micklebot]

Whether I've looked at team corsi numbers or individual numbers I've noticed many of the top players are from the same teams ie: BOS, LA, Chicago and NJ

So my question is: do you think corsi is more of a coaching thing than a player thing? There's no way that it just so happens that all the best corsi players are on certain teams.

3 of those 4 teams are really really good teams. I think overall team strength will boost your individual players Corsi. Now in the case of NJD, I don't know what to say. They play a very low instance brand of hockey, so I would imagine that they are much more susceptible to poor goal-tending and puck luck affecting the final results. A bad goal or bounce and they just can't make up for it like a high instance brand of play would. NJ seem like a good bet to surprise particularly if Schneider delivers.

 

[The Mighty Duck Man]

It goes both ways, somewhat. I think a lot of it is the systems those teams play, which explains why just about all the leaders belong to those teams and why just about every player on those teams has high corsi numbers. At the same time, though, not just any team can pick up and start playing a heavy puck possession style. The Leafs receive a ton of flack for their poor possession numbers, but they'd probably be murdered if they tried it. So, while part of it is coaching, I imagine a lot of coaches would go that way if they possessed the talent to do so. It's hard to say.

For instance, dumping it in is heavily criticized on these parts, and for good reason, but at the same time, skating it in is a lot easier to do when you're Patty Kane. And even for the rest of the NHL who isn't Patty Kane, if that's your skillset, that's the way to go about it. If it's not, it's probably not a good idea.

I'm not sure any coaching study will help a whole lot. As I alluded to earlier, a lot of coaches would probably play that style if they could, and you can really only work with what you have. Mike Babcock is a great example of that, as his Mighty Ducks teams were heavy defensive and not a possession team in the slightest. Fast forward to Detroit, and they're the ultimate possession team. Good coaches adapt, and more importantly implement what's going to work best with the group they're given, for the most part.

 

[TheDevilMadeMe]

Corsi is shot attempt plus/minus. Like regular plus/minus, every player on the ice gets a "plus" when there is a shot attempt for and a "minus" when their is a shot attempt against.

So of course regular linemates will tend to have similar Corsi ratings.

 

[JoelWarlord]

Possession can be very much driven by systems. Look at the difference between the Wilson Leafs and Carlyle Leafs in possession, or the 2012-13 aggressive forecheck and quick passing Montreal Canadiens to the 2013-14 dump and chase Montreal Canadiens. Similar rosters but vastly different possession games.

This is why Corsi Rel is used to compare players across different teams. Shawn Thornton is a better Corsi player than Christian Ehrhoff. However, he's 11th last in the league in Corsi Rel, while Ehrhoff is 44th. Corsi is a better stat to use on a team level. It's still relevant on an individual level but it would be stupid to call Shawn Thornton better than Christian Ehrhoff because his Corsi is better. Analyzing individual players needs to be done in combination with looking at the player's usage (zone starts, Corsi rel, QoC, linemates, etc).

 

[The Mighty Duck Man]

Possession can be very much driven by systems. Look at the difference between the Wilson Leafs and Carlyle Leafs in possession, or the 2012-13 aggressive forecheck and quick passing Montreal Canadiens to the 2013-14 dump and chase Montreal Canadiens. Similar rosters but vastly different possession games.

This is why Corsi Rel is used to compare players across different teams. Shawn Thornton is a better Corsi player than Christian Ehrhoff. However, he's 11th last in the league in Corsi Rel, while Ehrhoff is 44th. Corsi is a better stat to use on a team level. It's still relevant on an individual level but it would be stupid to call Shawn Thornton better than Christian Ehrhoff because his Corsi is better. Analyzing individual players needs to be done in combination with looking at the player's usage (zone starts, Corsi rel, QoC, linemates, etc).

I've really begun to like Corsi rel, but it too has it's drawbacks. A good, not great, player can have a superb Corsi rel because the rest of his team's terrible. On the flip side, a guy can have a terrible Corsi rel because he isn't much of a possession-style player and he plays on a possession powerhouse. Overall, no doubt it's a better measure than just individual Corsi IMO.

 

[GKJ]

This is from 2012 with regard to the system change from Terry Murray to Darryl Sutter

http://nhlnumbers.com/2012/5/14/lak-turn-around

 

[BroadwayJay*]

I've really begun to like Corsi rel, but it too has it's drawbacks. A good, not great, player can have a superb Corsi rel because the rest of his team's terrible. On the flip side, a guy can have a terrible Corsi rel because he isn't much of a possession-style player and he plays on a possession powerhouse. Overall, no doubt it's a better measure than just individual Corsi IMO.

Can you give an example of "good, not great, player" that has a superb Corsi rel on a team that is terrible?

It is still a team stat so a player with a superb corsi rel is probably doing a lot of things right, especially if his teammates struggle without him.

As far as a "non-possession-style" player, well isn't that an indication of what we're looking for? If he's isn't a possession-style player, as you say, and his relative corsi is poor; isn't that telling you exactly what it should?

 

[Feed Me A Stray Cat]

The Devils possess the puck a lot due to their aggressive forechecking system, which is the primary reason for their high Corsi.

 

[BigRig4]

3 of those 4 teams are really really good teams. I think overall team strength will boost your individual players Corsi. Now in the case of NJD, I don't know what to say. They play a very low instance brand of hockey, so I would imagine that they are much more susceptible to poor goal-tending and puck luck affecting the final results. A bad goal or bounce and they just can't make up for it like a high instance brand of play would. NJ seem like a good bet to surprise particularly if Schneider delivers.

The answer to the Devils is actually alot more simple than that. 0-18 in the shootout. They played wonderfully during the actual game. If they had even just a bad shootout record instead of historically terrible one, they're in the playoffs and probably pretty dangerous considering they wouldn't have to face a shootout.

 

[jc17]

This is from 2012 with regard to the system change from Terry Murray to Darryl Sutter

http://nhlnumbers.com/2012/5/14/lak-turn-around

Nice article. And looking at Corsi numbers, they did go up significantly after the coaching change.

 

[The Mighty Duck Man]

Can you give an example of "good, not great, player" that has a superb Corsi rel on a team that is terrible?

It is still a team stat so a player with a superb corsi rel is probably doing a lot of things right, especially if his teammates struggle without him.

As far as a "non-possession-style" player, well isn't that an indication of what we're looking for? If he's isn't a possession-style player, as you say, and his relative corsi is poor; isn't that telling you exactly what it should?

A guy I had in mind was TJ Brodie. Ridiculous Corsi rel for a guy who IMO is a good, but not great player, and would have a much more modest CF% and Corsi rel on a better team. Other guys in the top 50 from last year that I'd feel the same about are Mikael Backlund, Christian Ehrhoff, Jake Gardiner and Mike Cammalleri.

As for the other thing, I don't quite follow what you're saying. My only point is you might have a guy with an absolute dreadful Corsi rel, but all in all he's actually a pretty decent player, just playing on a possession powerhouse and looking terrible when it comes to those numbers, perhaps a guy who contributes in other ways. The Bruins fourth line from last year, which occupies 3 of the bottom 12 Corsi rel spots, comes to mind as an example of this. I don't believe they're that bad, but much different style-wise from the rest of their team.

 

[BroadwayJay*]

A guy I had in mind was TJ Brodie. Ridiculous Corsi rel for a guy who IMO is a good, but not great player, and would have a much more modest CF% and Corsi rel on a better team. Other guys in the top 50 from last year that I'd feel the same about are Mikael Backlund, Christian Ehrhoff, Jake Gardiner and Mike Cammalleri.

As for the other thing, I don't quite follow what you're saying. My only point is you might have a guy with an absolute dreadful Corsi rel, but all in all he's actually a pretty decent player, just playing on a possession powerhouse and looking terrible when it comes to those numbers, perhaps a guy who contributes in other ways. The Bruins fourth line from last year, which occupies 3 of the bottom 12 Corsi rel spots, comes to mind as an example of this. I don't believe they're that bad, but much different style-wise from the rest of their team.

It sounds like you just disagree with the conclusions we typically draw from Corsi and Corsi relative.

The point isn't that the Bruins fourth line is bad, it is that the Bruins fourth line is bad at possession. That doesn't necessarily mean that they are bad players. If they're instructed to dump and chase every ****, this is going to happen. If they can't complete zone entries because they lack skill, this is going to happen.

Why is it that you think TJ Brodie isn't good? I think it is extremely impressive to put up that Corsi rel on a bad team and I think ti tells us quite a bit about what a good possession player he is.

As far as having a more modest corsi relative, I think you might be right about that; but why is that a bad thing?

I often find that folks desperately want a "single solution" stat-wise, like Overall in hockey videogames. If you can take yourself out of that mindset (and I admit it is hard) you can really use corsi and corsi rel to give you a great understanding of who is valuable and who isn't.

 

[The Mighty Duck Man]

It sounds like you just disagree with the conclusions we typically draw from Corsi and Corsi relative.

The point isn't that the Bruins fourth line is bad, it is that the Bruins fourth line is bad at possession. That doesn't necessarily mean that they are bad players. If they're instructed to dump and chase every ****, this is going to happen. If they can't complete zone entries because they lack skill, this is going to happen.

Why is it that you think TJ Brodie isn't good? I think it is extremely impressive to put up that Corsi rel on a bad team and I think ti tells us quite a bit about what a good possession player he is.

As far as having a more modest corsi relative, I think you might be right about that; but why is that a bad thing?

I often find that folks desperately want a "single solution" stat-wise, like Overall in hockey videogames. If you can take yourself out of that mindset (and I admit it is hard) you can really use corsi and corsi rel to give you a great understanding of who is valuable and who isn't.

I probably should've clarified, but in a world where people are increasingly blurring the line between "good" and "good at possession", I was following the same idea. Too many people look at a poor Corsi or Corsi rel and say "well, that guy sucks", which just isn't true IMO.

There's also a simple logic to what I'm saying. The one thing about Corsi rel is that it doesn't measure how good at possession a player is, just how far ahead or behind he is compared to his teammates. So, in theory, absolutely you can have a guy who isn't necessarily good or bad possession-wise, but just so different from his teammates that it gives a strong Corsi rel rating either way.

As for Brodie, where did I say he wasn't good? I said he was a prime example of a "good, not great player" with a strong Corsi rel on a bad team. Again, I think on a better team, his possession stats would be a lot more modest. And there's nothing bad about that for Brodie, but it's just an example where Corsi rel can be skewed by great possession teams and bad ones. Again, I really like the stat, but just pointing out one thing to keep in mind when using it.

 

[Yossarian54]

Whether I've looked at team corsi numbers or individual numbers I've noticed many of the top players are from the same teams ie: BOS, LA, Chicago and NJ

So my question is: do you think corsi is more of a coaching thing than a player thing? There's no way that it just so happens that all the best corsi players are on certain teams.

I don't think it's more of either, in the sense that you could definitively tease it out at this stage.

Have a look at Tyler Dellow's stuff (http://www.mc79hockey.com/), this last year he's been on a bit of a quest to try and tease out coaching and player related Corsi events with particular attention to Taylor Hall, the Oilers and Bruce Boudreau. Dig through the last 10 pages of articles and you might find some interesting stuff. I think he's nailed down some definite coaching effects with regards to stuff Anaheim is doing, and also about Hall's zone entries.

 

[Hivemind]

Another interesting data point. A lot of uncontrolled variables in play, though.

 

[DTMAboutHeart]

It's kind of a mixed bag. Their are some teams who are so stacked with talent it would be pretty hard for them to not be good possession teams. Then theirs just bad teams who can't be saved. Their are a few direct cases of coaches/systems, playing a role:

NJ -- Posts consistently good possession while playing low event hockey but the results haven't quite followed. Definitely seems to be some weird systems thing.

TOR -- No other team is hurt more by their coach and his tactics. Possession continues to get worse even after bringing in solid possession players.

FLA -- Switched coaches during the year and saw their possession game improve accordingly.

 

[charlie1]

Another interesting data point. A lot of uncontrolled variables in play, though.

Very cool. Do you have a source for that, or did you make it yourself? Interested in reading the accompanying article if it exists.

 

[Hivemind]

Very cool. Do you have a source for that, or did you make it yourself? Interested in reading the accompanying article if it exists.

Credit goes to Japers' Rink. I pulled that one from their facebook page, and they had another, earlier version (Ovechkin era instead of McPhee era) on their twitter. I don't believe they've used either in an official article, to the best of my knowledge.

 

[KaseMeOutside]

i think the kings do something in particular that's very smart.

if you check their stats on extra skater, every player has a 50%+ offensive zone start rate.

now notice something else about the kings? elite faceoff team, if they have nothing going they have no problem throwing it at the net and starting clean, knowing that there is a good chance they will win the draws and can start a fresh offensive attack.

i think it also shows that they emphasize the goalie not freezing pucks and keeping the play moving as much as possible

 

[jc17]

i think the kings do something in particular that's very smart.

if you check their stats on extra skater, every player has a 50%+ offensive zone start rate.

now notice something else about the kings? elite faceoff team, if they have nothing going they have no problem throwing it at the net and starting clean, knowing that there is a good chance they will win the draws and can start a fresh offensive attack.

i think it also shows that they emphasize the goalie not freezing pucks and keeping the play moving as much as possible

You're right, I think that's huge. The top 3 teams in O-zone faceoffs were LA, Boston and Detroit while the top 3 in D-zone faceoffs were Toronto, Buffalo, and Edmonton.

 

[Bear of Bad News]

i think the kings do something in particular that's very smart.

if you check their stats on extra skater, every player has a 50%+ offensive zone start rate.

now notice something else about the kings? elite faceoff team, if they have nothing going they have no problem throwing it at the net and starting clean, knowing that there is a good chance they will win the draws and can start a fresh offensive attack.

i think it also shows that they emphasize the goalie not freezing pucks and keeping the play moving as much as possible

This is pretty interesting - if directing rebounds in favorable locations is in your goaltender's skill set (and avoiding a faceoff in general), then it's a strategy that could work very well.

The alternative (to defensive zone faceoffs or favorable rebounds) is unfavorable rebounds, of course, but Quick's low shot totals suggest that he's not doing that. We could probably estimate that more rigorously with play-by-play data looking for shots on net following a rebound.

 

[KaseMeOutside]

You're right, I think that's huge. The top 3 teams in O-zone faceoffs were LA, Boston and Detroit while the top 3 in D-zone faceoffs were Toronto, Buffalo, and Edmonton.

This is pretty interesting - if directing rebounds in favorable locations is in your goaltender's skill set (and avoiding a faceoff in general), then it's a strategy that could work very well.

The alternative (to defensive zone faceoffs or favorable rebounds) is unfavorable rebounds, of course, but Quick's low shot totals suggest that he's not doing that. We could probably estimate that more rigorously with play-by-play data looking for shots on net following a rebound.

I would be very interested in seeing team stats for icings per game, and the situations in which they occur, if you force another team into icing the puck you are creating a more favorable offensive situation than simply the goalie just stopping play(due to being able to create a line mismatch against tired players)

and look at the leafs, they are one of the worst faceoff teams in the league, yet also took the most. the devils were also bad, but they took over 600 fewer, did the leafs fail to recognize that weakness, or were they simply such a terrible team at possessing the puck that there was nothing they could do?

 

[Hivemind]

Tyler Dellow did quite a few articles on faceoffs (specifically faceoff losses and Bruce Boudreau) late in the season.

http://www.mc79hockey.com/2014/04/bruce-boudreau-and-offensive-zone-faceoff-losses/
http://www.mc79hockey.com/2014/04/a-quick-look-at-leafsducks-oz-faceoff-losses/
http://www.mc79hockey.com/2014/04/evolution-oz-faceoff-losses-in-2013-14/
http://www.mc79hockey.com/2014/04/pyrrhic-faceoff-wins-in-edmonton/
http://www.mc79hockey.com/2014/07/pyhrric-faceoff-wins-and-refining-problems/

The best general summation comes from the third link:

I’ve found that there are specific periods of time following a faceoff in which we kind of see the same results over and over. It’s 37 seconds for OZ faceoff wins/DZ faceoff losses, 29 seconds for neutral zone wins/losses and 21 seconds for defensive zone wins/offensive zone losses. That’s how it long it takes for the impact of the faceoff’s location and result to wash out of the results on the ice. Anything that falls outside of the period in which a faceoff impacts it, I call it an open play event and classify it separately.

The results are, year after year, incredibly similar.

That being said, do read all the links. Each has relevant and interesting information.


Aside of that, he also had a nugget in another article regarding zone starts, and players' impact on them (though the focus was on the defensive zone, rather than the offensive).
http://www.mc79hockey.com/2014/07/orpik-niskanen-joe-morgan-and-the-engine-room/

Zonestarts are a bit of a bugaboo of mine because I don’t think that they measure precisely what people think that they are measuring. In my experience – and I’ve been guilty of this too – people talk about them as if they are entirely a product of the coach’s decisions. I don’t think that this is quite right. The name is sort of misleading. If Johnny Bad Defenceman gets put out for a defensive zone draw and then ices the puck twice in a shift, he gets credited with three defensive zone faceoffs. The problem with that is that it’s not entirely a reflection of the coach’s decisions – it’s also a reflection of what he did when he was on the ice.

I wanted to test out this idea, so I gathered three years of faceoffs where Matt Niskanen or Brooks Orpik were on the ice and then checked whether or not they were on the ice for the last event in the NHL PBP files prior to the faceoff. What this does is let me calculate two different types of zonestart – the zonestart that is the true coach’s choice and the one that occurs during a shift.

There’s something interesting here, I think. It’s true that Niskanen gets slightly easier zonestarts than Orpik does when there’s a line change before the draw. It’s not a huge gap though: Niskanen’s got a 3.2 point edge. When they’re staying on the ice for a draw, the gap grows to 4.9 points. That suggests to me that at least part of the reason for Orpik’s zonestart deficit is…Orpik.

I'd love to see more studies in these areas, particularly regarding player impact on zone starts, as well as team impact on opponent's zone starts (specifically event driven regarding forechecks and icing).