Professor Fancystats Presents: Corsi

Whether you like it or not it’s a fact of life in 2015 that analytics, or advanced stats, have started to take hold of the conversation in all circles of hockey. For those of us who are not well versed in anything mathematics related understanding these concepts can seem fairly daunting. But fear not, our old friend Professor Fancystats is here to help guide you into the exciting world of hockey statistics! Our first post in an on-going series will talk about all things Corsi. Please enjoy and we hope this helps you get a better understanding of how these statistics work, and make you sound smarter at dinner parties or work functions!  

      • Orm Nullman

        Thanks for the link JW it was an interesting read. I was aware of much of the background of the quote I used, but not that Courtney became President of the Royal Statistical Society. The reason I used the quote was because Professor Fancystats in the video states that we use Corsi “to help approximate puck possession”. As noted earlier in the video Corsi was originally used to measure the work load of a goalie which I believe is a valid use. In my opinion it is also a measure of which team made more attempts to score but it is a poor proxy for puck possession.

        For example The Soviet Team of the 1972 Summit Series was a puck possession team that often circled back in to their own zone when their breakout play was obstructed thus maintaining puck possession. Also they only took shots at net when there was a high chance of scoring preferring to pass and retain the puck rather than shoot and possibly lose puck possession. As such I would expect that a Corsi for this team would be relatively low and a poor proxy for puck possession when in reality, if directly measured, their time of possession would be quite high.

        If you are interested in puck possession why not measure it directly rather than rely on a proxy. It was for this reason that I used the “Lies, damned lies and statistics” quote. Since using Corsi as an approximation for puck possession, as suggested by Professor Fancystats, is IMHO a poor use of this statistic and could be misleading.

  • Gordie Wayne

    Corsi? Why don’t they just call it “Pancakes”, makes about as much sense as Corsi…

    The NHL actually “renamed” Corsi when they included it on their website. It is now simply: Shot Attempts

  • Fancy stats are fun for fans, if you go over to Lowetides blog you will see that is all they use. I believe it is fun for them however I do wonder if they actually watch hockey.

    It has become the only tool in the tool box, as much as fancy stat supporters argue it is not the only tool, it is all they use.

    • BDH

      Man did you hit the nail on the head. The thing about advanced stats is that they are…advanced! People can’t just says “corsi” to everything. Its one tiny blip of information. There’s no point talking Corsi without also talking WOWY, zone starts, quality competition, PPM and more. But that blog is the laziest for reducing everything to corsi. The more I learn about advanced stats, the more I learn that there is a ton more to it than the average fan, me included, understands.

    • Reg Dunlop

      I was wondering the same thing. I actually missed some pretentious ass insulting average hockey fans, you know the type, hiding behind a keyboard while tossing insults that would earn him a whole lot of grief on city streets or a jobsite. And then I got to comment 42.

  • How could you not know that just by watching game ? If all the stats tell me I am not watching the game properly , then I simply do not bother to watch that sport any further . I enjoy it much more , and see more , with a minimum of stats to be honest .

  • You can’t keep track of what happens in a game. Stats are ridiculous. Common hockey knowledge says the team that tries the hardest always wins. You can tell Ryan Jones is better than Sidney Crosby, which is why “Goals” and “Assists” are bs.

  • Zarny

    Nice vid!

    The funny thing is there is no real debate.

    Nobody sane questions the concept of possession. To win you need to score goals and to score goals you need the puck. No coach in history has stated getting out-shot 45-18 and letting the other team play with the puck 65% of the game is the key to success.

    It’s not a coincidence many “old-school” tenets of the game about forechecking, backchecking, “winning battles”, picking up your man, etc, etc drive possession.

    The only hang up seems to be expressing it mathematically and simpletons who suggest they are being told to only look at a spreadsheet or that Corsi is all that matters.

    • fran huckzky

      Totally agree. Eakins believed that if you shot the puck enough times from the perimeter you would have great advanced stats and eventually you might even get a goal. Eakins was all about looking good. All sizzle,no steak.

  • Orm Nullman

    Loved the presentation of the vid, as well as the concept.

    I need to preface the following because I know I’ll get roasted: I’m a huge believer in statistical analysis. Resulting numerical models from statistical analysis are pretty damn accurate.

    But what I’ve always had trouble was the leap from “a shot on net reflects longer puck possession.” To me that seems too much of a leap in reasoning because of the many contradictions to that statement.

    Look at the Eakins PP for example, the Oilers could pass it around the perimeter, holding possession for lengthy periods of time, without even taking one shot.

    To me, a far more useful stat would be to literally count the time that a team has the puck on their stick. That would be a more accurate representation of “puck possession”. With that, I’d track recovered rebounds (with own and opposing goalie separately recorded) and stolen pucks.

    For me, I value shot attempts because it represents a better chance of goals being scored, for many reasons. From the perspective of “the physics of having a puck going in one direction increases the odds of a goal” to “goalies get sloppy and make more mistakes the more exhausted they are” to “an extremely high percentage of rebound shots are goals”.

    So I’d like to see what sort of evidence ties the number of shots to the time the puck is on a team’s stick.

    I’d also like to see why we look at shot attempts and not number of quality chances, when we look at using one metric to define a successful hockey club.

    Personally, I see “fancy stats” in its infancy. We aren’t recording enough information at the moment. Because of that we have very general models that I feel could be refined to be more specific.

    I was one of those that argued that the Flames high shooting percentage was by design, and I predicted that their shooting success would carry through the season. I think a large part of this is that they do not view 2 shots as equal. Some shots are better than others, resulting in a higher success rate per shot. Some shots were done purposefully not to score, but to put the goalie out of position, and to capitalize on the resulting scenario. I think this form of thinking comes about when one realizes that no two shots are equal. That high quality chances can be manufactured.