Quote:
Originally Posted by Jay Random
For instance:
http://rinkstats.blogspot.ca/2013/10...key-games.html
Money quote:
What surprised me is that the statistic that I would have guess correlates most strongly with winning (shots on goal) is highly correlated with winning, but in the wrong direction. That's to say, the team that takes more shots in a game is, on average, less likely to win the game. |
Yeah, so, I have no idea who that guy is, but he doesn't adequately adjust for score effects. "Games tied entering the third" often don't stay that way and score effects exist as early as the five minute mark of the first in some games. If he just removed all data from non-score-close situations, this would make more sense. Either way, it's pretty clear this dude has no idea what he's talking about - look at his conclusion; "win the faceoff battle". If you want to know how useless faceoffs are, I'll direct you here:
http://statsportsconsulting.com/main...lysis12-12.pdf
If you want to see the correlation and reliability for numerous different statistics as they pertain to winning, it's all right here:
http://www.pensionplanpuppets.com/20...-blocked-shots
Quote:
|
Now remember, Fenwick and Corsi measure shots on goal (or shots attempted), NOT puck possession. We actually have no good numbers for puck possession. If SOG are negatively correlated with winning, that pretty much eliminates the possibility that Corsi or Fenwick could be positively correlated. I suspect that may be too much to say based on the analysis given, but it certainly calls into question the idea that there is a strong correlation.
|
They measure shot attempts, there is no "or" about it. They are an excellent proxy for puck possession, which was determined when a bunch of leafs fans took a stopwatch and compared actual time of possession over a series of games in 2013 and found the results basically mirrored corsi. Another couple of guys did it with the Flyers, same thing. That being said, once sportVU or whatever is implemented and actual minutes / seconds of possession time are determined, these stats will basically be obsolete.
Hahaha, wait, your argument is that a team that scores more goals than their opponents is more likely to have won more games? Yes, if at the end of the season you look at the teams that have scored the most goals and allowed the fewest, you're likely to see them at the top of the standings. That's more or less a truism. What exactly is this supposed to show? The purpose of predictive analysis is to try to determine who will outscore who in the future.
Past goal differential does not do as good a job of this as possession (particularly fenwick close) - basically because of reliability. Refer to the pensionplanpuppets article above.
As for why I'm not willing to have the debate at any length (even this is a waste of time) it's that regardless, people with inherent biases will try to find reasons to disbelieve something that suggests their hockey team isn't very good. All of this information is out there and has been for more than half a decade, yet you still get people essentially saying "this is false because there are teams in the current standings that frustrate your predictions" like Indes, as if that demonstrated anything important. I could post every relevant article and answer every criticism (as has been done a thousand times on dozens of internet fora) and the same criticisms would arise a week later, expressed as though only just thought of for the first time. It's pointless.
Hockey fans exhibit some incredible tribalism, especially on the internet. As a result, I'd rather just say, "here's the data", and if you'd prefer to ignore it, fine.