Kukla's Korner

The Puck Stops Here

Steve Simmons Comment On Corsi

Steve Simmons writes a bunch of soundbites and publishes it as a column in the Toronto Sun each week.  Here is his most recent example.  In it one of his soundbites is

Another reason why I have next to great difficulty for the CORSI analytic statistic in hockey. I saw a team adjusted CORSI ranking for this past season. Tyler Seguin of the Bruins was rated fifth best in the NHL. The same Seguin whom Bruins coach Claude Julien kept on the third line, moving rookie Carl Soderberg from press box to first-line centre when Patrice Bergeron got hurt. I’ll take Julien’s instincts over strange numbers anytime

My first reaction when I read this was perhaps he read my work.  After all I published the top 20 players in 2013 by team adjusted Corsi this week.  That is unlikely to be a coincidence is it?  If I am the source that Simmons used, he definitely did not understand what he read and probably made no attempt to understand it.  That is easily shown because Tyler Seguin is not fifth best in the NHL, he is actually sixth.  That isn't a big difference and perhaps implies that Simmons saw some other list that team adjusted players in another manner.  In a worst case it shows Simmons was not very good at fact-checking.  However, whether Seguin is 5th or 6th really makes no difference to the point he is attempting to make.

The point Simmons is attempting to make is that if I scan a list of the top finishers in a given stat and there is one person near the top that I do not agree with, then the statistic makes no sense.  For example he could argue that he has big troubles with goals being used as a hockey statistic because Jiri Tlusty finished tied for fifth in goals in the NHL last season and there is no possible argument one could make that Tlusty is the fifth best player in hockey.  In fact I couldn't make a coherent argument for him being in the top 50 players in hockey.  Is that not the same point Simmons tries to make?  Does anyone seriously think it invalidates goals as a hockey statistic?  What it shows us it that goals (or team adjusted Corsi) or any other single number does not by itself rank NHL players.  In the subset of games we saw last year Tlusty was one of the top goal scorers in the NHL and Tyler Seguin was the sixth best player in terms of the puck possession that his team had while he was on the ice relative to his teammates.

If I as trying to make Simmons simplistic point that one player near the top of a given statistic invalidates the statistic, I would have found another player to pick on.  Why not Justin Williams of the Los Angeles Kings who appears third (ahead of Seguin) and is probably not seen as a player with Seguin's value (would you think Williams could have been traded for Loui Eriksson?)

With any statistic there are two groups of players who will excel.  There are the players who have significant skill in the area that the statistic measures and there are players who play in circumstances that maximize their ability to do well in the area the statistic measures (more realistically all players have a bit of both behind their numbers).  In order to "discredit" a statistic Steve Simmons style, one needs to identify a player who played under circumstances that maximized their ability to excel in a given area measured by a particular stat far more than his natural skills put him near the top in that stat.  You need to find someone like Jiri Tlusty in the goals statistic.

The problem is Steve Simmons didn't do a particularly good job selecting this player in Tyler Seguin.  Seguin is a very high draft pick (second overall) who has done well in terms of puck possession numbers or in terms of points per minute of ice time throughout his young career but has never been given a chance to play a significant frontline role.  Simmons criticism implies that Claude Julien is infallable and could never have kept a talented player from getting the chance he should have had.  This is a somewhat silly claim as hockey history is littered with talented players who were never given much of a shot with their first NHL team and moved on to another team to reach their highest levels of success.  Martin St Louis in Calgary, Dominik Hasek in Chicago and Marcel Dionne in Detroit are three obvious examples off the top of my head.  Some players were not given shots on well-run teams that won Stanley Cups including both Esposito brothers (Phil in Chicago and Tony in Montreal).  There are countless other examples we could find if we keep looking.  The possibility that Claude Julien was wrong in not giving Seguin a bigger role in Boston and eventually in getting him traded is quite plausible. 

Seguin was hurt in terms of total ice time because of the way he was used in Boston.  If he played more, his numbers would have been better.  However as a big fish on the small pond of the Boston third line Seguin was sheltered from playing in the toughest situations and that likely helped him in a statistic like a team adjusted Corsi ranking. 

The Seguin trade to Dallas was Tyler Seguin, Rich Peverley and Ryan Button to Dallas from Boston for Loui Eriksson, Reilly Smith, Joe Morrow and Matt Fraser.  Seguin, Peverley and Eriksson are the proven NHLers here.  Eriksson is probably the most established so far, as he appeared in the 2011 All Star Game.  The other four players are prospects and the prospects going to Boston are more likely to make significant NHL impacts.  The reaction from most NHL observers is that Dallas got the biggest jewel in Seguin.  He is a young player who is full of talent and looks ready to soon establish himself as an NHL star.  It is a big price for Dallas to have paid to obtain Rich Peverley and Ryan Button if Seguin was not regarded as a soon-to-be star player.  This is the kind of deal which could go a long way to rebuild Dallas and go down as one of the more lopsided trades in recent history. 

Seguin is a poor choice for Simmons to key onto to make his attempted point about Corsi.  In a couple years somebody merely posting Simmons words verbatim could lead to laughter.  I imagine there is a further agenda to bash Tyler Seguin or to suck up to Claude Julien. 

The biggest problem with Simmons writing is that he writes a series of soundbites and we are to guess at their context.  It appears he mistrusts Corsi largely because it is new and he doesn't get it.  He would probably have the same reaction if he was just introduced to goals as a new statistic and be amazed that somebody like Jiri Tlusty could be at the top.  Instead he picks a player who is regarded as a soon-to-be star because he wasn't given a frontline role in Boston as his example of a top team adjusted Corsi player who invalidates the statistic.  If Seguin makes the leap forward that many expect his point will look really silly.  Even if he doesn't, Seguin's puck possession success can be understood as a talented player gets to play a protected third line role allowing him to be a big fish in a small pond.  The biggest thing Steve Simmons needs to do to make a valid point is to stop writing a column of soundbites and better address any one of the points he thinks he is making in a soundbite.

Filed in: | The Puck Stops Here | Permalink
 

Comments

Avatar

I’m curious as to why you would say Marcel Dionne didn’t get much of a shot with his first NHL team (Detroit). A quick look at hockey-reference.com tells me Dionne led the team in scoring 3 of the 4 seasons in Detroit (second the other year), had 40 goals twice, and had 366 points in 309 games.

Posted by mc keeper on 08/25/13 at 10:20 PM ET

Avatar

So, Simmons must have read your stupid article and the evidence is that he got it wrong?

My god, you are some kind of psychopath.

Posted by Garth on 08/25/13 at 10:48 PM ET

PuckStopsHere's avatar

Dionne was god in Detroit buy rose to a much higher level in Los Angeles, but to some degree I am guilty of a “soundbite” that I didn’t back up at all.  if you don’t accept Dionne, try Cam Neely or Billy Smith instead,

Posted by PuckStopsHere on 08/25/13 at 10:49 PM ET

shazam88's avatar

lol you crack me up, Garth.

As for Marcel, while none of this matters in the least, some actually thought that it wasn’t a good trade from LA’s standpoint at the time:
“For Terry Harper and Dan Maloney, the LA Kings get baloney.”
True sign.

Posted by shazam88 from SoCal on 08/25/13 at 11:29 PM ET

DocF's avatar

Quoting Garth:  “So, Simmons must have read your stupid article and the evidence is that he got it wrong?

My god, you are some kind of psychopath.”

Yup.

The fact of the matter is that one can attempt to subject hockey to a statistical review, but it makes little sense.  The game is even less structured that basketball, for example.  A reasonably astute fan who has been watching the game for 20 years or more can probably evaluate a player better than can a whole hard drive full of statistics.  Individual performance is so tied to what one’s teammates do on the ice that it is folly to attempt to posit a statistical evaluation of a player.

Doc

Posted by DocF from Now: Lynn Haven, FL; was Reidsville, NC on 08/25/13 at 11:43 PM ET

PuckStopsHere's avatar

Doc

Since you say :

The fact of the matter is that one can attempt to subject hockey to a statistical review, but it makes little sense.

, I would love to see you back up that “fact” which is clearly an opinion and a questionable opinion at best.

Posted by PuckStopsHere on 08/26/13 at 12:54 AM ET

DocF's avatar

Puck head, read “How To Lie With Statistics”.  Your opinions were proven worthless in 2011 when you deleted every one of my posts that disagreed with your stupid use of unproven statistics.

Psychopath may not be an accurate description, but how does egomaniac sound?

Posted by DocF from Now: Lynn Haven, FL; was Reidsville, NC on 08/26/13 at 08:33 AM ET

PuckStopsHere's avatar

Doc

Your comments are useless wrong opinions that you repeated despite the fact they were shown to be false repeatedly.  Just like they would be this time if you would put up one piece of evidence to try to support yourself.

Posted by PuckStopsHere on 08/26/13 at 10:15 AM ET

Nathan's avatar

DocF, you’re missing the same point that Simmons is missing. The main folks that are painting Corsi (or any stat) as a catchall way to measure the greatness of a player are the ones that are criticizing the stat. To the folks that have worked to make mathematical adjustments to the data to enhance its value, and debated legitimate ways to better quantitatively measure different abilities of players, Corsi (or any stat) has never been a catch-all that should be trusted with 100% accuracy (or anything near it).

Long story short, you’re making a straw man argument by mis-representing the real position of the community that puts effort into developing more advanced metrics as a way to understand the game.

I’m all for the eyeball test. There is no doubt that in some cases, nothing can beat it. I would never argue that players should be judged strictly on some numbers or an aggregation thereof. After all, these are humans, and their play can be impacted by things that are difficult (if not impossible) to measure (emotions, basically). The whole point of discussing Corsi is to continue to pursue if the number has serious value in evaluating both teams and players, and then to use experiences in finding the pros and cons in Corsi to help think of new and better ways to objectively measure teams and players.

Criticism of any metric is welcome. The problem with Simmons’ argument (and your argument) is that the criticism doesn’t stand up to any amount of logical scrutiny. Simmons’ argument is based on a fundamental misunderstanding of what Corsi is intended to show. Your argument is, well, not really even an argument—you make a few statements hypothesizing that an experienced viewer can judge teams and players better than Corsi can, and that it is a folly to attempt to measure what Corsi tries to measure. But you haven’t even made an argument to support your hypotheses.

FWIW, you can go over to the post about the top players in Corsi adjusted for team and see that I am actually there, in the comments, being fairly critical of the stat as currently used. I’m not here defending the stat specifically. The difference is, I believe my critiques were constructive—I was trying to think of ways to improve what Corsi attempts quantify—and that my criticisms showed that I understood the purpose of the stat and that I made actual arguments, didn’t just hypothesize that it was wrong without explaining why.

Posted by Nathan from the scoresheet! on 08/26/13 at 10:35 AM ET

DocF's avatar

The arguments posited by those who feel statistics are a be all and end all (can you say MBA, boys and girls) are fine and dandy.  The problem is that not everything can be solved by plugging numbers into a formula and reading the result.  The reason so many businesses are messed up is exactly that.  MBAs are taught to plug numbers into a formula and read the result.  They do not ask if the formula is correct, if the numbers are accurate, nor can they determine if the result makes any sense.

Now if one studies accident statistics at a particular intersection to see if a traffic light would reduce the accident rate or if the lanes need to be reconfigured, that makes sense as these things can be quantified in a statistical manner. 

To determine if a baseball player is better than another can be aided by statistics as there is a large base of experience with this data and the game is more affected by individual performers and less by team performance.

Using a statistical analysis in a team sport which depends so much on team play is not very helpful.  Let us look at football, for an example.  Joe Smith is the running back for Highland Cream State Teachers College.  He averaged on 1.1 yards per carry.  Yet the line that he is running behind specializes in ole blocks.  That’s where the lineman waves at his opponent as he blasts by him to tackle the runner.  Can we tell how good Joe is from this performance?  Not really.  An experienced pro scout could look at Joe and interpret his performance to indicate that he could be an effective running back if he was playing behind a good line.  While the analysis gives us some facts, it never paints the whole picture.

Now I know I can never convince true-believers of the futility of CORSI numbers and I am not going to try any more.  Show me a coach who goes by the numbers and I will show you a loser.

Posted by DocF from Now: Lynn Haven, FL; was Reidsville, NC on 08/26/13 at 12:34 PM ET

J.J. from Kansas's avatar

Posted by DocF on 08/26/13 at 12:34 PM ET

Your whole argument isn’t an argument against statistical analysis in hockey. It’s actually the same argument with which the analytics crowd suffers:

Bad stats are bad and a bad understanding of what you’re studying is worthless.

The same problem ports the other way too:

Somebody who sucks at the eyeball test shouldn’t be using it either.  Would you believe that Joe Smith from HCSTC was the next Walter Payton if the janitor told you he was?

Calling data futile is the same as calling watching the game futile. It’s entirely wrong and it needlessly separates the two sides.

Now if one studies accident statistics at a particular intersection to see if a traffic light would reduce the accident rate or if the lanes need to be reconfigured, that makes sense as these things can be quantified in a statistical manner.

This is what hockey statistics aims to do and separating them out to just Corsi and saying that’s all they’re doing is like saying you’re only studying how fast cars are going in the accidents. Those wouldn’t tell you the whole story. The idea that Corsi is all that gets looked at is a strawman.

Hilariously, you actually can’t quantify how many fewer accidents you’d get at that intersection until you’ve put the model into practice.  All you can do before then is reasonably estimate a range based on a very large set of similar data.

That’s actually what statistical hockey analysis aims to do: they generally try to create a range of probabilities based on the study of similar data.  It’s not always right on, but just like accident rates are not a fixed number, they can get pretty close.

Posted by J.J. from Kansas on 08/26/13 at 12:56 PM ET

Avatar

Posted by DocF on 08/26/13 at 12:34 PM ET

You’re damning the hockey stats people with guilt-by-association of a group they’re not associated with. The leading people in this field tend to have master’s degrees in things like mathematics or electrical engineering (which requires a lot of math), not business administration.

This stuff’s a tool just like any other tool. One measure of identifying a player’s worth, not the only measure. Not necessarily even the best measure (I’ll take Chuck Grillo or Hakan Anderson’s eyes over Fenwick close any day of the week).

And as far as hockey executives who make use of “advanced” statistics, look no further than Stan Bowman, whose team just took home the Stanley cup.

Posted by larry on 08/26/13 at 04:41 PM ET

LiteWork's avatar

And as far as hockey executives who make use of “advanced” statistics, look no further than Stan Bowman, whose team just took home the Stanley cup.

Posted by larry on 08/26/13 at 04:41 PM ET

But how did advanced statistics help Stan Bowman?

Posted by LiteWork on 08/26/13 at 06:39 PM ET

PuckStopsHere's avatar

But how did advanced statistics help Stan Bowman?

The answer to this question is not in the public domain.  I am sure sabermetrics (a term I like better than advanced statistics - because I imagine the next wave of statistics to be super advanced stats and then super doper advanced ...) informed on all of the Blackhawk decisions but to different levels on different decisions.

If the question is intended on a lower level to be how do you know Stan Bowman really uses sabermetrics?  Here is the 2013 Hockey Analytics panel at the MIT Sloan Sports Analytics conference webpage.  Notice Stan Bowman is the only NHL GM on the panel.

Posted by PuckStopsHere on 08/26/13 at 09:12 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 The Puck Stops Here

imageThe Puck Stops Here was founded during the 2004/05 lockout as a place to rant about hockey. The original site contains over 1000 posts, some of which were also published on FoxSports.com.

Who am I? A diehard hockey fan.

Why am I blogging? I want to.

Why are you reading it? ???

Email: y2kfhl@hotmail.com