Throughout the year, I’m hoping to use my ELO ratings system to predict results and discuss various stories that emerge in the AFL season. I’ve described the model in a fair bit of detail elsewhere on the site but basically, an ELO rating system allows to compare the relative ‘skill’ of two teams and, using the difference in ratings, predict an outcome. It is nice due to its simplicity - the data it uses is the match margin and which team is at home, ignoring other information such as players, coaches, weather and so on.
Granted I’m writing this after round 5 and a fair bit has happened in the AFL already (such as Freo, Port, Collingwood and Richmond all sucking) but I wanted to have this post as a record of what my model thought at the start of the year. These ratings essentially are our end of year ratings for 2016, regressed toward the mean rating of 1500.
I may also update this post later with some simulations for the season but this is enough for now!