As more and more people get comfortable in using numbers to talk and think about hockey, I think it’s important to occasionally step back and think about what the numbers we use are actually telling us. Let’s use Ben Eager as a discussion point. I have my own calculation that I do of what I call an open play Corsi%. I wipe out the faceoff effects based on some math that I’ve done as to how long they persist and look just at what happened during the time in which there wasn’t a faceoff effect.
Here are Eager’s numbers since 2007-08, when data became available.
I’ve highlighted the good years in yellow and the bad ones in blue. You don’t need to be a rocket scientist to recognize that when Eager played on Chicago or San Jose, two really good teams, he was part of top notch fourth lines. I’ve talked about this before but if you’ve got a fourth line Corsiing (it’s a word) at 50%+, you’ve almost certainly got a hell of a hockey team.
In the summer of 2010, Eager left Chicago as part of the great blowup of the Blackhawks. Goes to Atlanta, who are terrible and waiting for someone to release them from their misery. A half a season of terrible Corsi% ensues. Atlanta, for whatever reason, decides that enough is enough and ships him off to San Jose. Poof! Part of a very useful fourth line as the Sharks go to the conference finals and Eager has what I consider to be his greatest moment.
Then he comes to Edmonton and posts horrible numbers again. His 2011-12 looked a lot like his 2010-11 (Atlanta edition) and then his 2012-13 came along and it was even worse and by the end he was living in OKC and it looked like his NHL career was over. The door may have opened a crack for him this year, with Craig MacTavish and Dallas Eakins kind of generally suggesting that slates are clean. MacTavish, in particular, had some kind words based on Eager’s play in OKC – but if he doesn’t play another game in the NHL, nobody would be shocked.
I can’t be the only person to whom this all seems a bit bizarre. If Corsi% is a measure of a player’s possession ability, how can it change so much if a guy switches teams? The answer seems kind of obvious to me: Corsi% isn’t a measure of a player’s possession ability. It’s a measure of what his team did when he was on the ice. We infer from that a guy has a certain set of abilities but, as Ben Eager shows, that’s not necessarily the case. There are a lot of other things going on that can impact on the Corsi% a guy puts up.
I’ve been rolling this around in my brain for a while now. I kind of think of this stuff as being sort of levels of a pyramid. The top level is Stanley Cups. The level below that is wins. Below that you have goal difference. Then below that, you have shots for and against (and Corsi) and shooting/save percentage. It’s basically a series of levels of things that we’re trying to connect to the level above it, if that makes sense. We figure out how one level works and then we try and figure out the level below it and how it connects.
Right now, people are kind of poking around in that level below shots for/against and shooting/save percentage, trying to understand what drives them. That’s where Eric Tulsky’s work on zone entries comes from. I kind of think that this is exciting stuff because it’s trying to answer the question that I raised above: what drives Corsi%? Once that’s figured out, you can start to do pretty cool stuff in terms of really isolating players who tilt the ice.
With that in mind, there’s some additional data on Eager that I thought was kind of interesting. I’ve been fooling around with isolating the number of “shifts” on which a guy allows one shot, two shots etc. “Shifts” is in quotes because it’s not really shifts – I group shots within a sixty second window together. What this basically does is take the ability to generate/prevent a first shot out of the picture and focuses on the ability to generate/prevent additional shots.
I think it’s useful because it kind of isolates part of the game. It takes the ability to gain the zone and generate that first shot out of the equation and focuses solely on the issue of generating/preventing additional shots. In effect, this is a sort of step towards breaking Corsi% down into components that measures efficiency at different things and isolating individual player contributions.
The data I’m presenting on Eager is only for “open play.” I want to do an apples to apples comparison here. What I’ve done is calculated how many Shot Attempts For (SAF) Eager’s team would generate per 100 shifts with at least 1 SAF with him on the ice. Then I did the same for Shot Attempts Against (SAA). I call it SAF/100 and SAA/100. What I found is pretty interesting.
2008 is a bit of a washout – Eager simply did not play very much. After that, run your eye down the SAF/100 column. Good Corsi% or bad, things look pretty similar. With SAA/100, it’s not at all the same. In his good years, Eager’s SAA/100 was under 140. In his bad years, he’s around 150+.
One wonders how much of that is on Eager and how much of that is on the players around him. It seems quite a coincidence that that number spikes when he plays on a poor team and settles down when he’s on a good one. It also seems reasonable to think that there are aspects of play over which certain players are going to have less and more impact when they’re on the ice. Take winning pucks on dump-ins for example. I don’t expect that the Oilers defence are going to have much in the way of an influence over how many pucks the Oilers win off dump-ins. That said, it’s going to show up in their Corsi% – a team that wins more pucks from dump-ins will have a better Corsi%.
One final table. These are the ratios of shifts with at least one SAF to one SAA for Ben Eager throughout his career. Again: when he’s posted a good Corsi%, the ratio has been above 1. When it’s been poor, it’s been below 1. How much of this is on him and how much is on the teams he’s played for, in terms of tactics and teammates? I’m not sure. What does he influence and what do his teammates/coaches influence? I’m not sure.
These are the questions that need to be answered though. Until we come up with definitive ways of breaking this down (and I believe it will happen), we should probably be a little cautious in terms of using Corsi% as a tell-all number. It’s absolutely an important indicator because elite teams control the game at 5v5 but it requires a little thought in terms of application.Email Tyler Dellow at email@example.com