Is corsi real?

[Mensch]

What I mean by this is how legit is this measuring “tool”? Do you think it has changed the game? Will it continue to change the game or will it fade out?

Also, help me understand corsi, fenwick and the other main advanced stats “tools”. I feel like I am struggling to understand these “tools” and feel like maybe I am losing interest in hockey or the nhl. But I know that knowledge is power so if I can learn and understand what all of this new stuff means and combine it with my “eye test” I think it would help me get back to enjoying the game like I once did. I used to feel like I knew so much but now when I talk on message boards or social media or with my friends I feel a bit out of the loop and am afraid to voice my opinions lol.

Click to read more...

 

[Doctor No]

For the "help me understand corsi" portion of your post, I highly recommend the stickied thread at the top of the forum:

http://hfboards.mandatory.com/threa...-by-the-numbers-forum-please-read-me.1585305/

Specifically the links in the fourth post (I didn't check all of the links to see if they still work, but I did check the Corsi/Fenwick one, and it does).

 

[Michael Farkas]

Understand history of "advanced" statistics. Individual player shot counts, time on ice (1930's), plus/minus (1960's), efficiency/save pct. (1980's), etc. all "advanced" stats at the time. Big boom, then as understanding accumulated, it was filed down a bit. Same thing with Corsi (originally, a measurement of goaltender work...goalie requires a lot of mental energy and physical energy, so even shots that go just wide or are blocked require you to be in your stance, puck tracking and even save selection), before the statistics community had enough seasons of data to draw meaningful conclusions, meaningful conclusions were drawn.

We've seen the limitations over the years - some quicker than others. Shot counting is biased against passive style defenders (those that yield the blue line), defensive/matchup centers, among others. It is biased towards low hockey sense, offensive defensemen who shoot from everywhere regardless of what's happening (this is not a desirable player at any level...Cody Franson is a good example) and streak-and-score wingers who have limited skills outside of skating and shooting and are not relied upon for defensive assignments.

The data is now public as most NHL teams and even high-resource major junior clubs have more useful, nuanced proprietary data - has a better pulse on the game. It has certainly not "changed" the game, it's just another tool and needs to be weighted appropriately. Like things like plus/minus and save pct., we'll look back years later and go "can you believe we..." and then whatever limitation associated with Corsi that's appropriate will be inserted in place of the ellipsis. Taking this statistic or any single statistic as a picture-painter is always a dangerous game.

 

[Mensch]

Thank you.

 

[Howie Hodge]

Not only is he real; he's got the most wonderful eyebrows! (Almost Eugene Levy class....)

 

[Howie Hodge]

Even his goalie mask didn't hide Jim Corsi's unique look.....

 

[morehockeystats]

Corsi is susceptible to Goodhart's Law, this is a huge drawback.

 

[Doctor No]

Agreed about Goodhart's Law. Once people know that you're measuring Corsi and that "MORE SHOTS = GOOD!!!!1111oneoneone", then people start taking shots for shots' sake, and not for the sake of winning games.

Corsi is 100% real, and it measures exactly what it claims to measure. The extent to which it's useful as a predictive measure for other things is where it gets sticky.

 

[Breakers]

When you combine corsi with other specific measures then it becomes useful.

I look more at the shot attempts against combined with transition data.
(how good are d-man at getting the puck out of their zone)
Its no coincidence that defenseman struggle with certain partners, and the data shows that.

when you look at d-man who struggle with shot attempts against you find d-men who aren't very good and have been bought out, exposed in the entry draft or bottom pairing.

Bieksa
Sbisa
Girardi
Garrison
etc etc etc

 

[Michael Farkas]

how good are d-man at getting the puck out of their zone

Tracking that directly and not a proxy of that gets you (royal you) closer to the right track, as an example of what I alluded to above. Good point.

 

[Cap'n Flavour]

Corsi is susceptible to Goodhart's Law, this is a huge drawback.

Uh, how so? Are you suggesting that skaters are purposely making suboptimal shot attempts just to boost their Corsi? Do you think that teams are actually rewarding players for good Corsi (and which specific measure)? I see no evidence whatsoever for either of the scenarios. Even if they were true, what exactly is your suggestion? Should the league and fans stop measuring or paying any attention to Corsi for fear of influencing the way the game is played?

 

[Doctor No]

Uh, how so? Are you suggesting that skaters are purposely making suboptimal shot attempts just to boost their Corsi? Do you think that teams are actually rewarding players for good Corsi (and which specific measure)? I see no evidence whatsoever for either of the scenarios.

You have to admit the possibility that both are happening, however. An easy way to get evidence of this would be to see if Corsi is less predictive of future victories than it was before it was widely promulgated. (I haven't done this, nor am I suggesting that you do it.)

Even if they were true, what exactly is your suggestion? Should the league and fans stop measuring or paying any attention to Corsi for fear of influencing the way the game is played?

I don't think that there has to be a suggestion; it's an observation about one limitation of Corsi as a predictive measure.

 

[morehockeystats]

Uh, how so? Are you suggesting that skaters are purposely making suboptimal shot attempts just to boost their Corsi? Do you think that teams are actually rewarding players for good Corsi (and which specific measure)? I see no evidence whatsoever for either of the scenarios. Even if they were true, what exactly is your suggestion? Should the league and fans stop measuring or paying any attention to Corsi for fear of influencing the way the game is played?

The teams might not be rewarding. The player agents may think differently though.
The key is for example, to look for trends in the contract year for players. And while this might not be happening, it might in the future.

 

[Thirty One]

Yes, Virginia.

 

[Doctor No]

Even his goalie mask didn't hide Jim Corsi's unique look.....

View attachment 94185

I don't want to be that guy, but why is Jim Corsi's mustache coming out of Eddie Mio's mask?

 

[Channelcat]

Its utterly worthless. Not even a proxy for possession. Continue to study the game and don't let yourself be led astray.

 

[Doctor No]

Its utterly worthless. Not even a proxy for possession. Continue to study the game and don't let yourself be led astray.

It may be "worthless" to you, but it has uses.

I appreciate bombast as much as the next guy, but your response has about as much worth as you ascribe to Corsi.

 

[Howie Hodge]

I don't want to be that guy, but why is Jim Corsi's mustache coming out of Eddie Mio's mask?

Good Call Doctor!! My memory fails me! Eddie Mio #31, Corsi # 35!! I should have looked that up; cuz ya got me on that!!

#35 was pretty follicularly impressive too!!

 

[Doctor No]

No worries - easy mistake to make! Getty Images does it on a regular basis.

 

[AfroThunder396]

These types of stats are descriptions of play, not explanations of it.

You're not good because your Corsi is good. You're good because you do X, Y, and Z right, which is reflected in your Corsi. A lot of people who are anti-stats fail to grasp such nuance.

To suggest that simply taking a lot of shots will magically win you hockey games is of course nonsense. Rather, it should be thought of in the context of "what attributes do good teams generally have"?

-Good teams, generally, outscore their opponents at even strength.
-Teams that outscore their opponents at even strength, generally, outshoot their opponents at even strength.
-Teams that outshoot their opponents at even strength, generally, outchance their opponents at even strength.

Of course, there are other factors that influence the success of teams. Some, like goaltending and special teams, can be directly measured and weighed. Others, like resiliency and speed, are tougher to nail down. I find them to be most useful when evaluating players on the same team.

 

[overpass]

These types of stats are descriptions of play, not explanations of it.

You're not good because your Corsi is good. You're good because you do X, Y, and Z right, which is reflected in your Corsi. A lot of people who are anti-stats fail to grasp such nuance.

That’s the question, isn’t it. Are those X, Y, and Z plays being reflected in Corsi?

Not always. See the dynasty Edmonton Oilers. They were outshot in every season from 1983-84 through 1987-88. So their Corsi would have been very average. Yet they were also 150 games above 0.500, outscored opponents by over 500 goals, and won 4 of 5 Stanley Cups.

NHL.com - Stats

So if you are going to use Corsi you have to know when it “works” and when it doesn’t. Or in other words, when can you assume shooting percentage for and against will be league average? When do you regress shooting percentage for and against to something other than league average?

 

[Setec Astronomy]

Uh, how so? Are you suggesting that skaters are purposely making suboptimal shot attempts just to boost their Corsi? Do you think that teams are actually rewarding players for good Corsi (and which specific measure)? I see no evidence whatsoever for either of the scenarios. Even if they were true, what exactly is your suggestion? Should the league and fans stop measuring or paying any attention to Corsi for fear of influencing the way the game is played?

There are whispers that Beau Bennett was doing this in his only season with the Devils, but that he didn’t fool anyone. He had a good Corsi on a bad team, and stat nerds were shocked that the Devils didn’t even bother to give him a league minimum offer despite needing RWs. The Devils have their own stat people who probably don’t rely on Corsi either as others above have indicated.

 

[AfroThunder396]

Right, which is why I said that generally, good teams will follow such patterns. But of course, hockey a very dynamic and complex system, and so some aspects like speed or toughness can't be captured with Corsi.

You're certainly right that "vanilla" Corsi has some serious limitations, but I'm not sure how that conflicts with what I wrote above. Only a fool would speak in absolutes or claim that Corsi is some kind of all-encompassing magic bullet stat. Of course there are different ways to build a team; Corsi just gives you the blueprint for one such way.

The dynasty Oilers, for example, followed a game plan of "do whatever the hell we want because we have half a dozen hall of famers". Great plan, but not something you can often emulate. It's not like you can tell your guys to just go out there and shoot at 26% like Jari Kurri and everything will be alright.

 

[Setec Astronomy]

Right, which is why I said that generally, good teams will follow such patterns. But of course, hockey a very dynamic and complex system, and so some aspects like speed or toughness can't be captured with Corsi.

You're certainly right that "vanilla" Corsi has some serious limitations, but I'm not sure how that conflicts with what I wrote above. Only a fool would speak in absolutes or claim that Corsi is some kind of all-encompassing magic bullet stat. Of course there are different ways to build a team; Corsi just gives you the blueprint for one such way.

The dynasty Oilers, for example, followed a game plan of "do whatever the hell we want because we have half a dozen hall of famers". Great plan, but not something you can often emulate. It's not like you can tell your guys to just go out there and shoot at 26% like Jari Kurri and everything will be alright.

Trying to emulate anything the 80s Oilers did is kind of pointless since goaltending was pretty atrocious until the early 90s. You could take ok open shots and you could come away with a goal pretty often, where a goalie saves it 95 percent of the time today.

If we could transport Al Montoya back to the 80s, we’d think he was the greatest goalie ever.

 

[ESH]

In the end, Corsi is just another stat. It tells you exactly what it measures, shot attempts for/against a player's team while they are on the ice. It's silly to try and claim a player is better simply because of Corsi. That's not to say it isn't a useful stat, I'm just saying that stats can only tell you what they measure. It's influenced by all kinds of things such as zone starts, linemates, strength of competition, and team strategies. There's only so much that Corsi can tell you, and I think that problems arise when people try to use it as more than it is.