Is Corsi losing its Correlation with Winning?

I had someone e-mail me today asking me why some traditionally good Corsi hockey teams are struggling.

Any explanation for the NHL season thus far? I hate to use PDO, but is it just the year of PDO without much further explanation? I can’t believe how many traditionally sound possession teams are struggling this year, Dallas really surprised me and of course as a Blues fan seeing Chicago faring so well of course makes me nauseous.

I had been looking at some other stuff recently and I noticed that it seemed that there were fewer really good and fewer really bad Corsi teams this year. Anyway, this e-mail prompted me to look at this in more detail so I looked at team 5v5Close CF% over the past 5 seasons to see if there are any trends and see if there is in fact a tightening of the spread in CF%.

The first thing I looked at was to take the average of the top 5 CF% teams and the bottom 5 CF% teams and all the groups of 5 in between for each of the past 5 seasons and then plot them (I am using 5 team averages to smooth things out a little, particular at the extremes). Here is what I found.

As you can see, there is clearly a tightening in CF%, particularly for last season and this season. The worst 5 teams are not as bad and the best 5 teams are not as good. There has been some compression in the middle as well over the course of the 5 seasons.

Another way we can look at this is by looking at the standard deviation of team 5v5close CF%.

There has been a clear drop off in the past two seasons from ~3.8 down to ~2.2 this season.

So, what is the reason for this tightening spread in CF%? Well, two years is hardly enough to be sure of a trend but my theory is this is a result of analytics. Analytics put a big spotlight on Corsi. The hype probably hit its peak 3-4 years ago when teams (particularly bad teams) started hiring analysts from the analytics community. More and more teams are now putting more focus on Corsi and puck possession hockey and in turn there are fewer really bad puck possession teams which means the good teams are having a tougher time maintaining a higher CF% (no bad teams to beat up on). The result, the spread decreases.

I actually predicted this would happen in an article I wrote September 16, 2013. The article was about Corsi and not putting too much focus on it to the detriment of other important factors that contribute to what makes players and teams good.

In the article I discussed the evidence that teams prioritize shooting percentage (or skills that lead to elevated shooting percentage) to the detriment of Corsi/puck possession. The evidence was salaries seemed more tied to shooting percentage than Corsi.


So, if teams have been doling out contracts based on, effectively, shooting percentage then it is perfectly reasonable to assume that shooting percentage talent is more evenly distributed across teams than Corsi-talent is. Under these circumstances Corsi would be highly correlated with winning percentage because that is where the differences lie between teams.

This lead to two things.

First, shooting percentage skill was highly valued and thus was more evenly spread across all teams.

Second, Corsi/puck possession was an under valued skill so teams that focused on that skill were able to get a lot of it really cheaply.

Both of these skills are significant factors in what makes good players good but the result of the above scenario is that winning will be largely driven by Corsi/puck possession since that had a larger talent spread across teams.

When I wrote that article I had predicted that if teams put more focus on puck possession hockey the result would be that Corsi would become less correlated with winning and shooting percentage more correlated with winning.

Furthermore, if General Managers as a whole started paying primarily for Corsi we will start to find that Corsi talent becomes more evenly distributed across teams and thus shooting percentage would become much more highly correlated with winning

I think we are starting to see some of this and is leading to observations like the one my friend made and e-mailed me about. The top Corsi teams are not necessarily winning as much as top Corsi teams in the past, simply because the gap between them and the rest of the teams has closed. If all this is true, being a better final standings will be less tied to final Corsi standings now than they had in the past. This is not to say that puck possession is less important, just that there is less of a talent disparity across teams than in the past.

Ultimately, as I stated in my article from 3 and a half years ago, focusing on outscoring your opponent is the primary goal. It is just not so easy to accomplish that through out shooting your opponent as it was in the past.

This article has 2 Comments

  1. same thing happened to baseball with OBP and pretty much anything else that you can measure. observer effect in action — measuring something changes that thing.

Comments are closed.