Over the last number of games of the season, I’ve been tracking Zone Exits and Zone Entries Against by Oilers defencemen. If you’re unfamiliar with controlled zone exits and entries, please follow those links for background on them and my work thus far.
***
A few people have asked me how these zone tracking numbers (zone exits and entries) relate to corsi. The answer is that they certainly should, but I guess I don’t know for sure. My aim here is to take the games I’ve tracked so far and see how well they correlate. Obviously with a sample of 10 games and 7 defencemen, I’m not looking to prove some larger connection. Mostly, I’d like to satisfy my curiosity and see how my tracking numbers correlate with corsi in these last 10 games.
A Quick Primer on Corsi
“Corsi” is simply a measure of shot attempts. Shot attempts at your opponent’s net is “corsi for” (CF) while shot attempts against your own net is “corsi against” (CA). Your relative corsi (corsi-rel) is your ratio of shot attempts on the ice relative to when you’re off the ice. So if a player has a positive corsi-rel, their team does a better job of outshooting the opposition when they’re on the ice compared to when they’re off (and vice versa for a negative corsi-rel).
It’s pretty intuitive that having having a good corsi is a positive thing. If you’re getting more shot attempts than the other guys when you’re on the ice, that’s usually a good sign. It means you’re in the other team’s zone having meaningful possession.
Controlled Zone Entries Against – What do we expect to happen?
Eric Tulsky was the first (as far as I’m aware) to describe that carrying the puck in (vs dumping it in) leads to far more shot attempts. Jen Lute Costella also reported that teams that allow more controlled entries tend to give up more shots against them. This suggests that defencemen who allow more controlled zone entries against should allow more shots.
Now, I did not record every zone entry in the games I tracked. Since my goal was to try to identify a specific defensive skill, I only recorded zone entries where a defencemen was targeted. Entries where a forward was targeted or dump-ins from a long way back were not recorded. Further, if Clendening allowed a controlled zone entry on his side, his partner (Nurse) would not be knocked for that. So the Controlled Entries / 60 stat only includes entries that a particular defenceman is involved with.
Theory: Oilers defencemen who allow more controlled zone entries should have a worse corsi.
Controlled Entries Against – What did happen?
I took the 10 games that I have Zone Entry Against data for and got my friend @OilersNerdAlert to pull the corsi data for those games. (Side note: check out his fantastic blog: oilersnerdalert.wordpress.com!)
The chart below has the Controlled Entries Against / 60 and the CF-rel for each defenceman who played in those 10 games (ordered from best to worst for Controlled Entries).
Just glancing at the data, it looks like there’s a clear pattern. In fact, there is, with a correlation coefficient (R) of -0.90. (An ‘R’ of 1 or -1 implies a strong correlation while 0 implies no correlation.) Let’s look at it graphically.
That looks pretty good! In this 10 game sample, defencemen who allowed more controlled zone entries had a worse corsi-rel and the correlation was reasonably tight. The main outlier below the line-of-best-fit (meaning his corsi-rel was lower than expected) was Pardy (who only played 3 games in the sample). The largest outlier above the line (meaning his corsi-rel was higher than expected) was Oesterle, who was also excellent at Controlled Zone Exits, which may have been improving his corsi.
Let’s take this a step farther. Which is correlated with Controlled Entries Against: ‘corsi for’ or ‘corsi-against’?
Turns out… both! The correlation coefficient is in opposite directions but pretty close (-0.73 vs 0.83). In these 10 games, preventing controlled zone entries was correlated with a better corsi-for AND corsi-against.
It’s also reassuring to me that this matches up other data and serves as a bit of quality control. I was concerned with only tracking “targeted entries”, wondering if that was a bit of an artificial distinction. Next year, I’m also going to track ‘untargeted entries’. My theory is that those are mostly noise, but we’ll find out when I get the data.
Great… So What About Zone Exits?
The aforementioned Jen Lute Costella wrote (paraphrased): exiting the zone with control leads to going on the attack ~88% of the time while dumping the puck out leads to the other team going on the attack ~70% of the team. This suggests that defencemen who have more controlled zone exits should be rewarded with more shots going at the opposition’s net.
Theory: Oilers defencemen who have more controlled zone exits should have a better corsi.
Controlled Zone Exits – What Did Happen?
Let’s look at a similar chart, but this time for Controlled Zone Exits, once again ordered from best to worse for zone exits. Though I tracked 11 games, I only included the 10 games here that I also tracked Zone Entries Against to keep the corsi data consistent.
Hmmmm… that doesn’t look quite as strong by eye, especially with Mark Fayne sticking out with that positive corsi down at the bottom. Indeed, the correlation coefficient (R) of 0.38 implies a pretty low correlation… which is made abundantly clear when you see the graph…
That’s not nearly as strong and (despite the trendline) you could argue that it looks pretty random.
Why The Weak Correlation?
Well it’s only 10 games and 7 defencemen. There’s not enough data here to make any sweeping statements one way or the other. However, it does make some logical sense that the correlation would be weaker. Defencemen are only involved in a portion of all zone exits. There are plenty of zone exits carried out only by forwards. Those are not accounted for here but certainly impact corsi for and against. Meanwhile targeted Zone Entries Against appear to make up a majority of zone entries. Further, allowing a Controlled Zone Entry often leads directly to shots against, where as there are a couple steps in the way before you can get controlled possession in the offensive zone off a zone exit.
Somebody must have answered this question already. There are other much larger samples of zone exit data from previous seasons from various hockey teams, including the Oilers (put together by Bruce McCurdy). Of course, there’s also The All Three Zones Project put together by Cory Sznajder, where he tracked every game for every team in the 2013/14 season. I’m certain someone has looked at these larger sources of data and explored the correlation between zone exits and corsi, but I haven’t found a comprehensive post about it yet. Perhaps someone can point me in the direction of one.
I did find a post at Pension Plan Puppets where @draglikepull looked at an early sample of Sznajder’s work for the Maple Leafs defencemen, performing a similar analysis I just did but with 46 games of data. He found a reasonable correlation between Controlled Zone Exits/60 and corsi-for but not for corsi-against.
Summary
As of now, in the 10 games I tracked, having fewer Controlled Entries Against was clearly correlated with a better corsi. This matches what I would expect based on the work done by others.
However, having more Controlled Exits was only weakly correlated with a better corsi. This does not match what I would expect based on others’ work. I suspect this is due to sample size but it may be suggestive of a weaker link between zone exits and corsi than zone entries against.
I haven’t seen any public articles using The All Three Zones project (or another large zone exit project) and relating it corsi, though I’m sure they must exist. If they don’t (though it sounds overwhelming for my meek stats skills) I’ll try to tackle that question.
Next Step!
Acquire more data! I’m going back and tracking games from earlier in the year. I’ve done three already. I’ll try to update as I go along, focusing on individual defencemen (Klefbom, Davidson, Gryba).
Thanks for reading along!
Add The Sports Daily to your Google News Feed!
