I don't get this group. How is Florida State ranked, while Texas A&M and Clemson aren't? Never mind that Boston College is 4 spots behind. I didn't put them on my poll, which means that Florida State got an average of #19 from the other 3 voters. I don't know how you look at that team and still think the Seminoles are a top 20 team.
alabama is all lowercase, please don't fix it.
The spreadsheet of votes is locked. I can't see where all the polls ranked everyone.
I see a few other trends, but I think they are just outliers due to the small sample size. I bet it will all get figured out in the next 2 weeks.
Capitalization is for proper nouns and alabama is not a proper place...
FSU got ranked highly due to one of the computer polls, FSU has played more games but they'll drop out of that poll in the coming weeks (check the poll in this comment: https://lemmy.world/comment/12160842). But with only 4 ballots that's enough to keep them ranked.
I'll look into unlocking the poll results, I certainly want this to be transparent
I see now about Florida State. I remember that poll from last year was had a few outliers early on, but I felt like it came together nicely at the end. It explains almost all of their points.
I didn't rank Florida State, so make that 2 other voters. b/t this poll and last year's polls, some of the people using computer polls need to work on refining their algorithms :P
I too would like to see the spreadsheet.
Hopefully, there are more submissions in the coming weeks.
I see I'm catching a lot of flack here for my computer poll (and understandably so), to which I say: vote in the poll! The more participants, the more my funky algorithm gets drowned out. I will say, I did tweak it this year to further devalue preseason expectations, but I can't let it just stand on 1.5 weeks of data. Preseason expectations get fully flushed out after week 4.
Also, because the transfer portal makes it so hard to get a read on rosters early in the season, I outsourced the returning production variable to Action Network's metric. It worked pretty well last year, but I'm not sure they're on the mark this year. If anyone has a better source (that is free or cheap), I'm willing to give it a try for next year.
My ballot -- still a little weird from preseason expectations, but that's mostly factored out now. The bigger issue is that 2-3 weeks of data is too small a sample size for the computer to know what to do with. Teams that played P4 teams are very high (even if they lost), whereas teams that played FCS schools are very low (even if they blew them out). Teams that already had a bye are even lower (exacerbated by this season having an extra bye week). It should all be sorted out in a couple weeks.
I will say, by eye test I really disagree with some of these rankings (I mean, ou up to number 4???) but I find the pure data intriguing, especially for spots 40ish through 134 which a lot of people ignore. I'm considering a few tweaks (to diminish the effects of garbage time stat padding), but it would make the algorithm more complex. Without fully automating it, I need to keep it simple enough to tabulate semi-manually every week. Hoping I can use next offseason to tinker with some scripts.
So just checking, are we supposed to submit our ballots for week 3 in the link above? Because I did, but looking at the results it seems like it might still have week 2 ballots as well.
Yeah, my general workflow is honestly sloppy af with this thing. But I typically just leave the poll up as the same link and just clear the submissions every week. I do all the calcs on the sheets doc anyway (cut and paste the weeks out to their own sheets each week) so it doesn't really impact the results. Unfortunately I'm only ok at remembering to do all of this lol. It should be all up to date now.
Edit: I'm also likely to split the results and vote posts up again, as I imagine the other 2 voters last week probably want their name in the pot for the gift card. And it may be unclear the I need some sort of fedi user verification.