Re: Alaska / UAA : the Strategic, Restructured, Combined 2019-20 Season Thread.
The preseason polls are nothing more than a reposting of last year's tourney field with a few "big name" teams that missed the tourney sprinkled back in because the media can't not vote for them.
So I've spent a lot of time thinking about this, as well as some time playing with numbers. In general, past success presages future success; an Elo-style rating is generally useless as a predictive tool if you go back and say that everyone is equal at the start of the season. Even if you do a regression to the mean — which is what I've typically done — you can't regress too far to the mean and have value. How valuable? Stay tuned.
That said, one thing that is true is that an Elo-style rating would differ from the tournament field. Sometimes, a good team is going to miss the NCAAs given the size of the field. Sometimes, a team puts a big lead out early and fades at the end of the season. I remember a couple of years ago that BELOW was saying good things about Bemidji the season prior to their Bitzer-fueled heights, despite the fact that they'd been a sixth seed in the league the previous year.
Having been a USCHO poll voter, the other thing that I can say is this: I didn't get to watch many non-CHA games back in those days, as this was pre-streaming and non-regional TV coverage was largely limited to the postseason. You're ... you're just making stuff up, looking at standings, squinting at counting stats, and looking to see how many draft picks are on a team. I don't think that I ever spent more than a half-hour coming up with my votes, and while I'm sure that some folks were more responsible with their votes than that, the countless stories of "random SID intern" voting as "coaches" is real.
GFM
