Corsi, Shooting Percentage and Coaching Changes

On the weekend I posted an article looking at the relationship between Corsi and Shooting percentage and suggested that good Corsi teams are often poor Shooting Percentage teams and that there is generally a negative correlation between Corsi and Shooting percentage. This relationship seems to hold for most teams except for the elite teams or the truly bad teams. Yesterday over at I looked at this relationship just prior to, during, and just after the Randy Carlyle coaching era and it seemed to hold true (to some extent) for the Leafs during that time period.

These kind of relationships sometimes brings on a negative reaction among those familiar with Hockey Analytics and in particular those that believe strongly in possession and Corsi. I sometimes wonder why this is because we see this relationship occurring all the time with score effects and score effects is a well known and accepted concept in hockey analytics. Let’s recall what score effects are:

  • When a team is leading they will generally give up more shots and take fewer (resulting in a depressed Corsi) but generally the shots given up are of lower quality resulting in higher save percentage and the shots taken are of higher quality resulting in a higher shooting percentage.

So, due to some difference in playing style, when a team is leading they will see a drop in their Corsi and a boost in their shooting percentage. This is the exact same thing as the negative correlation I am observing in these articles. Why people find it hard to accept here but accept score effects is beyond me but some people have trouble with this. In any event, I want to take a look at how the relationship between Corsi (CF%) and shooting percentage has changed over the course of the season for the four teams that have made a coaching change thus far – Senators, Oilers, Devils and Leafs. Let’s look at these teams in reverse order and start with the Leafs first because I have already discussed them in the article and I’ll leave the Senators to last since they have the most interesting results. So, with that said, here is the 5v5 CF% vs Sh% chart for the Maple Leafs this season.


The black line indicates the time of the coaching change and what you see are the rolling averages over a 500 corsi event (for + against) sample. The correlation between these two is -0.20 so we do see a negative correlation. What we also see is that the Leafs CF% was actually rising under Carlyle prior to him being fired and the shooting percentage had already started falling off as well.

How about the New Jersey Devils?


The correlation between CF% and Sh% for the Devils is -0.38, or a fair bit stronger than for the Leafs. The Devils have been on a run of much improved shooting percentage recently but that has corresponded with the lowest CF% levels of the season. While Sh% seemed to be on the rise prior to the coaching change it did jump up a bit more after the coaching change though has dropped back the last little bit. Overall the highest shooting percentages on the season have occurred after the coaching change which is also when the Devils have had their worst CF%. Surprisingly, the Devils might be one of the worst possession teams in the league right now.

And the Oilers?


The negative correlation is quite strong here as the correlation coefficient is -0.795. Early in the season the Oilers had a low CF% and a higher shooting percentage which then reversed into a higher CF% and a lower shooting percentage before them both converged in the middle just prior to the coaching change. After the coaching change the Oilers CF% dropped to season lows while shooting percentage jumped back to early season highs (though it has fallen off in recent games).

For the Leafs, Devils and Oilers it is difficult to say that their coaching changes have had a major impact thus far (maybe for the Leafs but it is too early to tell) as it seems for all teams their post coaching change trends appear to have actually started just prior to the coaching change. Everything is different for the Senators.


Unlike the three other teams, the coaching change in Ottawa appears to have a significant positive impact as both their CF% and their shooting percentage has increased dramatically from where they were just prior to the coaching change. When you see stuff like this you really wonder if this is in fact one of those instances where the coach (in this case Paul MacLean) really did lose confidence of his players. The coaching change really did seem to have a positive impact on both CF% and shooting percentage. This surge in both CF% and shooting percentage means the two statistics are positively correlated over the course of the season with a correlation coefficient of 0.30.

In the future I’ll maybe take a look at a few other coaching changes from past seasons (i.e. Pittsburgh hiring Bylsma, Anaheim hiring Boudreau) to see how they looked and I might also take a look at save percentages as well. So far though all evidence points to the existence of a negative correlation between CF% and Sh% though there are also some exceptions to that rule like the Ottawa Senators after their coaching change.

This article has 5 Comments

  1. I agree that it is surprising the resistance to the idea of inverse correlation between Corsi For % and Shooting %, particularly since as you mention it is demonstrated empirically (the work in Hockey Abstract covers this quite thoroughly) and also makes sense intuitively, but I believe at least part of the resistance is coming from your use of the percentage share as opposed to looking at rate correlations.

    I am in general a strong believer at examining underlying rates as I believe they contain extra and important information beyond simply percentage shares (a shared opinion with @RegressedPDO, @IneffectiveMath, and @acthomasca). It is very different if a team has CF% of 55% with CF/60 55 and CA/60 45 as opposed to one with the same CF% but rates of CF/60 22 and CA/60 18. I think it would be much more reasonable to expect the first team to have their CF% negatively correlate with Sh% because of their high shot attempt generation whereas the second team has the same CF% due primarily to shot attempt suppression and would not necessarily be expected to have a negative correlation with Sh%.

    As to your last note of items for further investigation, I’d also like to see the impact on a team like the Capitals after Boudreau was fired and they brought Hunter on. Keep up the good work!

  2. So I looked at your earlier article and did a similar analysis of my own with some slight variations. I used data from War-On-Ice; Even Strength 5v5, All Score Situation – Score Adjusted data (as many have shown, and I concur with, using just Score Close is bad and Score Adjusted is better), looking at total team data from 09-10 season through the 13-14 season. I used rate information of various types and so compared CorsiFor Sh% (GF/CF) vs CF/60, FenwickFor Sh% (GF/FF) vs FF/60, and ScoringChancesFor Sh% (GF/SCF) vs SCF/60. [Scoring Chances is using the War-On-Ice definition which is available on their website] I then cut out the top-3 and bottom-3 teams measure by GF/60 (sticking with my rate methodology). These teams were CHI, PIT, BOS, and FLA, MIN, NJ.

    Interestingly, despite my variations in methodology I got similar results, the R^2 for the CF correlation was .2966, for FF it was .2698, and for SCF it was .3262, all with a solid negative relationship between production rates for and relevant Sh%. More precisely I think it speaks to the idea that teams that pile on the shots indiscriminately are doing so at lower levels of quality, be it a function of being behind more often than even score adjustment can account for or systems/coaching or simply the poorer decision making/play of the teams players. Here are some links for the pics of these three scatter plots:

    I went one step further with is and examined the other end of the table; for these same selection of teams, how do their CA, FA, and SCA rates compare to their relevant Sh%? If there is indeed a persistent and inverse relationship between shot quantity and shot quality then this should be observable at both ends of the ice. We would expect to see that teams giving up greater numbers of shot attempts should also see the % of those attempts going in to be less (this would be the defensive argument of shot quality, that “we allow more shots but they’re outside and lower%”, also know as the “Toronto Leafs strategy”). However, contrary to this expectation, the effect goes away completely. For CA, FA, and SCA rates there ends up being essentially no relationship between the rates against and the % of them that go in. The R^2s are CA at .0634, FA at .0599, and SCA at .0697. Why would this be? My best guess is the conclusion would be teams have the most control over their own behavior and results in terms of the quantity and the quality of that offense they generate but have little impact on the quality of offense directed against them, and can only distinctly impact the quantity of offense directed against them. I think this would be an important conclusion and supportive of the statements made by @RegressedPDO in terms of shot suppression being extremely important. Here are the 3 charts for the ‘against’ relationships:

    Also, just to try it out I rejiggered the against analysis to change which teams I excluded from the top and bottom to be based on best and worst in GA/60 rather than keeping the same teams from the GF/60. While I don’t think this approach is correct since it’s very much skewing the sample to fit the result you want to achieve, I was curious. (perhaps I should’ve just done both based on GF% but not sure that would change results much) Even when making this adjustment, the results don’t much with the R^2 for CA at .0786, FA at .122, and SCA being the only significant change moving to .3412 (oddly enough the strongest relationship out of all 9 metrics examined). This last point is very interesting, particularly in how it diverges from the Corsi and Fenwick measures. I would have to think more as to why they could be so different.

    1. Thanks for the comment(s). Definitely some good stuff here. I am a proponent of splitting up CA and CF from the percentage. With the dataset I was working with at the time I couldn’t calculate this but it was on my todo list.

      It is definitely more difficult to find defensive relationships. I have found this in nearly everything I have researched. I am not sure exactly why but I suspect a big portion of it is as you suggest, offense is more in your own control. I do think that some players are capable of suppressing shooting percentage against (boosting save percentage) but far fewer players are able to do that than can boost offensive shooting percentage.

      I also did some research that seemed to indicate that coaches are actually quite poor at identifying defensive players so maybe that is a part of it too.

  3. Great article, David. When you mentioned using a “500 corsi event sample”, do you mean your sample size was 500 games?

Comments are closed.