Quote:
Originally Posted by Itse
Arbitrary means quite a few things. It also means "based on individual discretion".
Just because someone says that "these are the important numbers" doesn't make it so. This is one of the major misunderstandings of the whole fancy stats community.
Just having a system is not the same as "not arbitrary", and even further from "objective fact".
Stats like "bad goals" can by definition only be a arbitrary because "bad" has no fixed meaning and it's not a thing you can measure. You have to just make up some definition of bad, which is probably not going to be much better than someone else's definition of bad.
"Most goals scored" is not an arbitrary stat. "Best goalscorer" is.
SV% is not an arbitrary stat, "bad goals" is.
|
There is certainly discretion in picking parameters and how to weight them. But that's hardly arbitrary.
As Enoch pointed out the definition basically eliminates what you're saying ...
Quote:
|
based on random choice or personal whim, rather than any reason or system
|
If they have a log of 32 stats there is certainly a system. There isn't any whim once they have built the model.
You can certainly disagree with the parameters used and the weighting, but to suggest it's just subjective arbitrary list wouldn't be fair.
Most "fancy" stats are pretty clear cut in what they are counting. Then they're pressure tested against actual out comes looking for the degree of causal relationship.
Most goals isn't a model. It's a summary. As we learned from 2018-19 to 2019-20, you couldn't project the Flames or their players simply by saying they did this last year, so that's the model.
Why? How? You have to look at circumstances beyond just a scoring summary.