03-20-2018, 12:49 AM
|
#210
|
|
Backup Goalie
Join Date: Apr 2013
Location: Calgary
Exp:  
|
Quote:
Originally Posted by Calgary4LIfe
The advanced metrics are what they are - just more information to digest.
What they are not is this predictive model that is often trotted out.
Flames under Hartley were horrible in most advanced metrics, and were pointed out as being 'unsustainable'. They kept that 'unsustainable' tag for 1.5 seasons (until the goalies crapped the bed, ran into some injuries, and had the second pairing defencemen suddenly regress horribly).
Flames under Gulutzan were advanced stat darlings and they were 'about to break out' all season long, but never did.
Just on one single team we have such polar opposites of what the advanced metrics were predicting. Winning under Hartley was unsustainable, but it was sustained long enough. Losing under Gulutzan was unsustainable, but was likewise sustained long enough.
The model is faulty. The analysis that people keep pushing doesn't have a confidence interval worth squat. It uses an indirect measure of possession (shots taken), and then goes on to explicitly state that it is indeed possession. It is not, and the confidence interval disproves any significant correlation IMO. Just on this one team you have results that are in complete disagreement.
Advanced metrics ARE useful, and they ARE interesting. The trouble with them is that they are incomplete, and you have nonsense stats like "PDO" that are 100% garbage (for heaven's sake, a good team with good offensive players, and with great defensive players and a strong goalie will have better than 100 AND IT IS NOT EXPECTED TO REGRESS TO THE MEAN in a season or even in the future - that's WHY they are a good team!).
Too many junk stats, too many inferences on metrics with too loose of causal relationships, and too many people shrieking about their predictive nature.
The Flames played for CORSI, and they were damn good at it all season. I hope Treliving has seen that strong CORSI should not be a system that you play for, but rather a natural by-product of having a good team playing well.
|
Goodhart's law strikes again - "When a measure becomes a target, it ceases to be a good measure." https://en.wikipedia.org/wiki/Goodhart%27s_law
Oh well some more games to get through as a fan and then maybe get out and enjoy summer
|
|
|