The Limits of Corsi: What can and can't it do?

[hatterson]

Found this article. Good refresher/informer on what corsi is and what its limits are.

http://www.arcticicehockey.com/2014...ts-evaluating-players-teams-hockey-statistics

Every few days, a person of some public renown mentions Corsi, which is closely followed by the entire hockey Twitterverse exploding into a brouhaha. People then proceed to squabble and bicker for a while until things eventually die down.

Rinse and repeat.

But what is Corsi? What is Corsi not? What can it and can't it do? What relationships does it have with players?

If you were to put it simply: it is a slice of the pie that is what makes a player or team effective but not the whole pie.

I think the author does a good job of explaining the origin, strengths and limits of corsi.

Click to read more...

 

[Bear of Bad News]

When my biggest complaint about an article is that I wished it had been longer, that's got to be a good thing.

Anyhow, I enjoyed this treatment - thanks for digging it up!

 

[TheDevilMadeMe]

This part is overly simplistic, and epitomizes my biggest problem with the majority of statistical analysises in hockey:

Corsi is not something that gets inflated simply by shooting more often.

Recently Jordan Eberle depicted (very accurately) the basics of Corsi. In the sound clip he mentions that when players do things correctly -like breakouts and zone entries- a player will naturally improve Corsi; he also states that once a player starts throwing the puck as soon as they cross the blue line, things will start to go south.

This is correct. Looking at the last three seasons combined, there is a relationship between a player's Corsi and their personal shot attempt rate, albeit a very weak R^2 of 0.27. Keep in mind that being in another team's zone more will naturally give more opportunities to shoot and that better players will be able to create relatively more as well.

Seems there is, in fact, a small correlation between shot attempts and a player's personal Corsi, but it is dismissed as insignificant. Whatever, fine. Seems like there is little if any relation between the average player's shooting and his Corsi. But what about the outliers - guys like David Clarkson who take bad angle shots all the time? Treating all players as similar datapoints and completely ignoring different styles of play.

And actually, that's my biggest problem with Corsi - it is usually applied to all players evenly. Yes, a puck moving defenseman will have a better Corsi than one whose skillset is best at keeping shots to the outside in the defensive zone, but that doesn't make him the more effective player necessarily.

 

[JoelWarlord]

And actually, that's my biggest problem with Corsi - it is usually applied to all players evenly. Yes, a puck moving defenseman will have a better Corsi than one whose skillset is best at keeping shots to the outside in the defensive zone, but that doesn't make him the more effective player necessarily.

As far as the handwaving of Corsi for the sake of Corsi, IMO it's because team's don't play to maximize their Corsi. The assumption (which I think is reasonable) is that a player won't just throw the puck away all the time, team's try to make scoring chances so it still works as a stat. Corsi has more utility as a team stat, it's not really fair to penalize a team's shutdown line for playing against the best players, and most people control for this. I don't agree that it's applied to all players evenly because good analysis uses QoC, zone starts, TOI, and other controls. Nobody worth listening to uses raw corsi to evaluate players. The other aspect is that it's the sum of all 5 players on the ice and the corsi events for and against. If a shutdown D lets a puck mover freewheel, it will show up in his corsi because he's on the ice while it's happening.

I think this is a very interesting part of the whole debate. I guess the question we need to ask is whether or not different styles of play are equal, is a top tier shutdown D equivalent to a top tier puck mover? I guess for me my view of the game is that the best defense is a good offense. If a guy's skillset is to keep shots to the outside it might lower the chances of a corsi event becoming a goal, but it drastically reduces the likelihood of his own team recording corsi events or shots (and goals).

We can look at the Karlsson vs Weber or Suter vs. Subban debates here, it's just hard to find that point where we're overvaluing a player's skill in his own zone over a player's ability to not be there in the first place. I think intuitively we all understand that a defenseman bringing offense is the better player (no matter how good Vlasic is defensively he won't be in the Norris talks) but the challenge comes from comparing guys at the lower end of the scale, the Gleason vs. Liles trade in Toronto comes to mind.

 

[Dream Big]

As far as the handwaving of Corsi for the sake of Corsi, IMO it's because team's don't play to maximize their Corsi. The assumption (which I think is reasonable) is that a player won't just throw the puck away all the time, team's try to make scoring chances so it still works as a stat. Corsi has more utility as a team stat, it's not really fair to penalize a team's shutdown line for playing against the best players, and most people control for this. I don't agree that it's applied to all players evenly because good analysis uses QoC, zone starts, TOI, and other controls. Nobody worth listening to uses raw corsi to evaluate players. The other aspect is that it's the sum of all 5 players on the ice and the corsi events for and against. If a shutdown D lets a puck mover freewheel, it will show up in his corsi because he's on the ice while it's happening.

I think this is a very interesting part of the whole debate. I guess the question we need to ask is whether or not different styles of play are equal, is a top tier shutdown D equivalent to a top tier puck mover? I guess for me my view of the game is that the best defense is a good offense. If a guy's skillset is to keep shots to the outside it might lower the chances of a corsi event becoming a goal, but it drastically reduces the likelihood of his own team recording corsi events or shots (and goals).

We can look at the Karlsson vs Weber or Suter vs. Subban debates here, it's just hard to find that point where we're overvaluing a player's skill in his own zone over a player's ability to not be there in the first place. I think intuitively we all understand that a defenseman bringing offense is the better player (no matter how good Vlasic is defensively he won't be in the Norris talks) but the challenge comes from comparing guys at the lower end of the scale, the Gleason vs. Liles trade in Toronto comes to mind.

I haven't yet got the knack of Corsi. Trusting all of you that do to enlighten me.

Question I have is how does the coach's style and instructions/control of how a defence man plays get caught/explained by Corsi. For example Caps current coach Oates and previous coach Boudreau. Completely different coaching styles have to impact Corsi.

 

[Yossarian54]

I haven't yet got the knack of Corsi. Trusting all of you that do to enlighten me.

Question I have is how does the coach's style and instructions/control of how a defence man plays get caught/explained by Corsi. For example Caps current coach Oates and previous coach Boudreau. Completely different coaching styles have to impact Corsi.

Have a read of these:

Not defenceman specific, but discussion of coaching tactics and corsi with relation to Boudreau and the Caps.

http://www.japersrink.com/2013/2/7/3961818/boudreau-ovechkin-capitals-trap-fml

http://www.mc79hockey.com/?p=6918