View Single Post
11-15-2012, 04:40 PM
Bear of Bad News
Mod Supervisor
Bear of Bad News's Avatar
Join Date: Sep 2005
Location: Windsor
Posts: 5,332
vCash: 663
Originally Posted by habsfanatics View Post
I tend to side with the guys who know the numbers better than I do, like yourself and Iainn and czech your math, but I see some valid arguments from Rhiessan that seemed dismissed out of hand.

I have seen throughout this thread that even guys that know the numbers better than I do and insist themselves that there are some flaws and then turn around and immediately post about so and so doing this and using AS as their only source of proof. Czech has done it several times now, even admitting to not seeing many of the seasons or games discussed prior to the lockout. This isn't meant to bash, but I see them being used as the be-all end-all quite often.

The only point I have and have no idea if it really has any merit is that it would seem likely that the distribution of points wouldn't remain static from top tier to bottom tier players. If the league gets tougher, it may not impact the skilled guys as much as the tweeners. Seems pretty logical, no?
If anyone - Czech or otherwise - is using adjusted stats as the alpha/omega of an argument, that's a problem.

Regarding your last paragraph, I agree completely - although if we're attempting to measure "value", then it raises different questions entirely.

The chief adjustment that adjusted statistics make is to adjust for different goal environments. If a goal is scored in a league where the average score is 3-2, that's a lot more valuable than a goal scored in a league where the average score is 8-6. If you're going to do that, then you probably need to take it a step further, and instead of adjusting to league-wide scoring levels, you need to adjust to a schedule-weighted scoring level (since games in the 1980s Smythe Division had less-valuable goals than games in the 1980s Norris Division, for instance).

Then do you go further, and adjust for situations? (A goal scored in a 1-1 tie game is more valuable than a goal scored at the end of an 8-1 win). And at what point does the sample size wash out the value in the information?

Adjusted statistics are never the ultimate result - but they add information, and they're generally more informative than raw numbers.

Bear of Bad News is offline   Reply With Quote