With the crazy AFL fixture starting tonight there is lots of potential for ladder changes. Here are my model's thoughts on the likely ladder at the end of it all.
(Round by round predictions here: https://t.co/LT6Yy7d63t)
@WATOsPod@HunterGMeredith@Barrel_Randall@sc_mag_aus @AFLFootyLive Actually been a late change to the model meaning it’s picking Geelong tonight (by half a point). Other picks the same but margins different
@CarltonFC@CarltonHistory 1. Kouta
2. Fev
3. Kouta’s last quarter in the 99 prelim
4. Marc Murphy (still a scapegoat for some despite years of loyalty, captaincy and awards)
We are all thinking the same thing... Hopefully, #AUSvNZ is a lot more competitive than #AUSvPAK, but @ocfitz1 has compiled the numbers to back up those hopes.
Join us as we crunch the numbers on two batting lineups that appear to be quite evenly matched
https://t.co/3Z6zNv86Q1
@HunterGMeredith @jacobjewson @sc_mag_aus It did pretty well in testing, MAE of 0.125. It got half the top 10 right last year so hopefully is decent this year
Here is your very own Brownlow tracker according to my model with predicted votes for each round for the predicted top 30. Something for fellow Brownlow nerds such as @jacobjewson to print off and use as the night goes on.
@HunterGMeredith @jacobjewson @sc_mag_aus Very basically, it’s a random forest model so a bit of a black box. But is based on player stats and game margins and trained on data from 2014
Building a model for the Brownlow medal, here are the correlations between player stats and Brownlow votes on a match by match basis since 2015. Obviously this is less useful for stats like hitouts where most players get zero each week and so correlation is low.
Had a look at the best batting averages in test history across eras. Whilst more players have averaged over 40 than ever before, Smith still stands out for his time (still not comparable to Bradman though). See link for interactivity: https://t.co/68FVtihq8d