Quote:
Originally Posted by Enoch Root
Is it? If it isn't accurately reflecting what it is attempting to measure, then it isn't a good model.
Not attacking you, just saying hockey stats still have a long way to go. And the argument 'they're the best we have" doesn't hold water if they aren't accurate and meaningful.
|
I think so. I linked to the rationale in the analysis thread. It's not perfect as I would prefer a positional adjustment rewarding centers and punishing wingers because of the difficulty required to play each position. I also think that I agree for the need to treat various game states differently, which is unique to hockey, but the results should be normalized to yield a (near) constant number of wins every season.
If "they're the best we have" doesn't hold water, then neither does anyone's eye test. You'd have to give a rationale explaining empirically why you believe so, which is next to impossible given that almost no one has time to watch every game.