Quote:
Originally Posted by Shazam
Take corsi. Now, it's a correlation, so right on that basis it should not be trusted.
|
I think this is the trap that every advanced stat advocate falls into. Corsi does not correlate very strongly with team success. If you want a predictive measure earlier in the season of where your team is going to finish, then
look at goal differential.
I do think Corsi is useful. But as I’ve detailed in the Tkachuk thread, I use it more for individual player assessment with respect to teammates, not use it to take the average of an entire team and trust it as an overarching performance indicator. It’s why I consider the whole “high danger”-ness of a corsi event such a useless, garbage metric.
Regular CF has a higher correlation to team success than HDCF over a bigger sample dataset... it’s a joke.