Didn't the Oilers specifically speak to 'playing for CORSI' a few seasons ago? I remember Eberle talking about it, and describing it a bit with what Eakins wanted. There was a lot of low perimeter shots, but you aren't necessarily throwing the puck away each time. As long as your team is skating well, you should pick up some of those loose pucks (the goalie isn't going swallow those pucks every time, even if they are low-percentage shots). Conversely, the other team may in turn have a tonne more possession and be way more of a threat to score, but they may not get off as many shots. Over-passing, missing the net, etc. I think there are times when you watch a team in the offensive zone and they look menacing, but sometimes they didn't even get a shot on net. Look at the PP for instance - even with the man advantage and ample zone time, the shot counts are often low for the PP. Sometimes they are really high, but sometimes they are low even with abundant possession time.
Eakins definitely got caught 'playing for CORSI' - unless you think Eberle (and I believe Hall as well, though I might be misremembering here) were flat-out lying. I remember Eberle specifically stating that they were told to shoot at the net regardless of how they felt about the stats, and that they were trying to get their CORSI numbers up.
Did the Flames do this? Hmm.. I don't think so. I do think that at times, their shot selection was questionable, but I do think this was more a symptom borne out of frustration more than an implementation of anything systemic.
I do think that Itse does have a point there. I do think that a team can indeed have a higher CORSI without actually having more actual possession. Does it happen often? I don't know, but that is the risk with counting stats that infer something. CORSI SHOULD directly align with possession, and now with the widespread and accepted use, it is synonymous with possession.
It would actually be interesting to see how closely that it actually aligns. My bet is that confidence rate here is pretty high with actual possession, but just from the simple fact that it is NOT directly measuring possession, there are bound to be instances where it is not only misleading, but also wrong.
Is the game too fast to really use stop watches to count possession like they do in soccer? I think you end up with ambiguous times as well as puck battles play into the equation, but you can eliminate those and other common plays from the counted time. Maybe use both CORSI and actual timed possession together?
There has to be instances where CORSI is not indicative of actual possession. We have seen it not be indicative of actual pressure at times. It is what it is - shot attempts that has a summary and an inference made that is (probably) very highly tied to actual possession. I would be really interested to see how closely those numbers follow an actual timed possession stat, however. If anything, it will help to prove or disprove certain notions. I would actually love for someone to go back in time and measure Edmonton's "playing for CORSI" and see how off it was.
I don't think CORSI is a garbage stat, but I also don't see it as infallible either. It is an indirect inference of possession by using a stat that SHOULD be higher with actual possession, so of course the less direct you are measuring something, the less accurate that measurement is going to be, right?
|