Welcome to 2026's hottest new model @HAH_Binnie! Now on the Squiggle platform. That's 21 models good to go, plus a handful either putting on the final touches or having crashed and burned in the off-season (not yet sure which).
Well done the Cats & all that but time we shouted out the real champions, @plusSixOneblog on Tips & @andrew_whelan on Bits & MAE. Too bloody good - well done!
@MatterOfStats Cold Sims are way easier to program and run so much faster if you do it right. Please find some evidence that they are better. I'm happy to ignore any evidence to the contrary until you do!
Has anyone seen a review of the efficacy of "hot" vs "cold" simulations (ie simulations where you update, say, Elo ratings within the sims based on simulated results for earlier games, versus those where you don't)?
I know FiveThirtyEight uses hot sims, but can't find any data
@SquiggleAFL@daveslutzkin Thanks for pointing this out - found a bug where the Gabba was listed as being in Victoria so the Lions got a penalty for playing interstate! Should be fixed for this week and hopefully explains most of my models poor performance this year!
@NickWelch51 I’m still unconvinced by season sims this early in the year (cc @MatterOfStats) BUT this round has caused some wild fluctuations in mine and Sydney now a 45% chance of top 4 and projected 2nd most wins plussixoneblog.com/page/aflm-pred…
@MatterOfStats@GRAFTRatings@SquiggleAFL The quality of forecasters coming through suggests that (at least for me) this streak might be coming to an end pretty soon. I think we need an official name before that happens so we can always look back on that as a feather in our caps!
@GRAFTRatings@plusSixOneblog@SquiggleAFL Just had a look, and it seems that the four of us are now the only forecasters to have been in s10 on Squiggle for all four years
I feel like we should come up with a name ...
Objective ranking of who I would like to see win the #AFL premiership, using a GUT based model
Brisbane
St Kilda
Bulldogs
West Coast
Power
Richmond
Collingwood
Geelong
@thehulksheroes Thanks mate. It’s using an R package called {reactable}. It does require some HTML to get it looking like it does but I just used their examples and modified from there glin.github.io/reactable/
@plusSixOneblog I have had a read through your blog and am frothing the Elo current tips/predictions tables. Is that custom JS you wrote, or was this a modified package in R?
Fellow @FMI_TL ers, please join me in offering huge thanks to FootyMaths for forging on and putting in all those hours despite the especially trying circumstances in this weirdest of years.
Not even a global pandemic could stop this competition.
Hip hip ...
AFL Wonk Tweeps on @SquiggleAFL (or others!) who only produce estimated Margins, how do you calculate ladder percentage? I just use `average margin` as a kind of surrogate but that falls down towards the end of the season where percentage matters