Kukla's Korner

Kukla's Korner Hockey

Corsi Number Gets No Love

from Bruce Dowbiggin at the Globe and Mail,

You have to admire Elliotte Friedman of Hockey Night In Canada. The man tries. Like Galileo attempting to get the Catholic Church onside, Friedman is singlehandedly trying to drag HNIC’s high priests into a hockey renaissance.

During one of Saturday’s segments, Friedman bravely resurrected the Corsi Number, a metric that measures how many shot attempts by his own team and the opponent a player is on the ice for. Many analysts see it as a far better stat than plus/ minus.

You’d have thought he’d just told the David Suzuki fan club there might be a slight hitch in Dave’s idea of imminent extinction via climate change. Panelists Glenn Healy, P.J. Stock and Kevin Weekes either mocked the metric or ignored it. The trio seemed more interested in the faux-couture collar on Stock’s jacket.

The disbelieving look on Friedman’s face was like that of a man watching three people trying to discover fire using wet sticks. If you want to try an office pool, guess the next time Friedman bothers mentioning a hockey metric for the red-meat chorus.

more on HNIC and some 'defensive' talk...

Filed in: NHL Media, Hockey Broadcasting, CBC HNIC, | KK Hockey | Permalink
 

Comments

Avatar

One would think that, of any of the “advanced” stats, Corsi would be one of the easiest for fans to understand and accept. There’s no math or arbitrary assumptions like those that go into GVT or “point/win shares” (however defined). There’s no complex averaging like qualcomp. The only thing you need to be able to do is count.

Sure, it’s an imperfect figure that must be understood in context, but anyone who thinks +/- is in any way meaningful should be a fan of Corsi, because all it is is basically a more accurate version of it with less noise. You still have the same problems with it being affected by roles and teammates, but you smooth out a lot of the “luck” problem to get a more useful number.

But I think part of the problem is that it (1) has a weird name that doesn’t describe the stat and (2) you either need to run your own script or go to a fan-run website like behindthenet that has such a script in order to get the data. Those things put people off. If it has a weird name and nobody ever talks about it and you can’t look it up on NHL.com, it must be some crazy maths cooked up by people who don’t understand the game.

If NHL.com would start publishing Corsi ratings on its player stats pages with a name like “shooting +/-” I’d bet you’d start to see more mainstream acceptance.

Posted by Sven22 from Grand Rapids on 03/25/13 at 06:27 PM ET

Savage Henry's avatar

The only thing you need to be able to do is count.

There’s your problem.  It’s out of reach of many in the hockey media.

Posted by Savage Henry on 03/25/13 at 09:42 PM ET

phillyd's avatar

Baloney, it’s a made up stat. How many “shots” are dump ins during a line change or floaters down the ice while killing a penalty, or even flips it at the goal at the end of a period. I’d probably say 4 or 5 a game. When the average number of shots is 30, well, you’re talking about 1/6 of all shots being pretty much throw away ones. That’s a significant piece. That doesn’t include floaters from the end boards. If you want an effective “stat” for player ratings, combine possession with shots from a “hot area” like the a geometric shape from the bottom of the crease to the bottom of the face-off circle to the top and across (think like a lane in basketball). The other issue with it is it will always adversely affect the players who’s primary job is to check the top line of the other team. They will always have a low value because they aren’t there to generate offense, and if they do, it’s a bonus. You also need to remove PP time for it to be effective. I’ve seen no indication that this is done (could be wrong, however) like it is with +/-.

Posted by phillyd from Southern New Jersey on 03/26/13 at 09:12 AM ET

redxblack's avatar

i"d much rather watch a game than count a game. The stats are less relevant to me.

Posted by redxblack from Akron Ohio on 03/26/13 at 10:08 AM ET

Avatar

When the average number of shots is 30, well, you’re talking about 1/6 of all shots being pretty much throw away ones. That’s a significant piece.

What percentage of goals are lucky bounces? The result of nothing positive or negative you actually did on the ice? The result of the goaltender blowing it? I’m not saying all shots are created equal but not all goals are either. Both have problems but using shots tends to work out better because you’re working from a much larger sample. And over a long enough period of time those throw-away dump in shots are going to affect player Corsis more or less equally, much faster than the fluke goals balance out.

If you want an effective “stat” for player ratings ...

That’s not really what Corsi is for. I would never ever argue that higher Corsi = better player or that it’s the “be all” hockey stat. It has lots of problems. All I’m saying is that it’s much better than +/- is at what +/- is supposed to do.

The other issue with it is it will always adversely affect the players who’s primary job is to check the top line of the other team. They will always have a low value because they aren’t there to generate offense, and if they do, it’s a bonus

Totally agree, but +/- has exactly the same problem, only worse for two reasons: (1) +/- is more affected by luck because the sample size is much lower and (2) Corsi is much better at taking the goalie out of the equation. Skaters and players share responsibility for goals, but shots are almost entirely controlled by skaters.

You also need to remove PP time for it to be effective.

Most of the time the number is reported for 5-on-5 play only.

AND FINALLY,

The real test of whether or not these numbers are useful are how well they predict future success or failure. On a team level, goal differential does a better job at predicting the future than team record, and Corsi does even better than goal differential.

Again, no one would argue it’s a perfect stat. Like any other stat it requires context, for all the reasons you describe and more. All I’m saying is that, while it’s imperfect, it’s better than a lot of the stats out there that get used without question.

There seems to be this feeling out there than unless a new stat is perfect it’s stupid and useless. I don’t get that. It’s like saying if we can’t end crime but we can reduce it by 40% we shouldn’t do it becuase it doesn’t fix the problem. I’m not saying that’s your perspective necessarily (I don’t want to put words in your mouth) but I get that feeling a lot when I talk about this topic.

Posted by Sven22 from Grand Rapids on 03/26/13 at 03:13 PM ET

Avatar

TL/DR version: Corsi is not a be-all stat, nor does it pretend to be. All it is is a demonstrably better way to do +/- with better predictive power, especially at the team level.

Posted by Sven22 from Grand Rapids on 03/26/13 at 03:22 PM ET

Add a Comment

Please limit embedded image or media size to 575 pixels wide.

Add your own avatar by joining Kukla's Korner, or logging in and uploading one in your member control panel.

Captchas bug you? Join KK or log in and you won't have to bother.

Smileys

Notify me of follow-up comments?

Feed

Most Recent Blog Posts

About Kukla's Korner Hockey

Paul Kukla founded Kukla’s Korner in 2005 and the site has since become the must-read site on the ‘net for all the latest happenings around the NHL.

From breaking news to in-depth stories around the league, KK Hockey is updated with fresh stories all day long and will bring you the latest news as quickly as possible.

Email Paul anytime at pk@kuklaskorner.com

 

image

image

image