Inspired by this in depth article from Footy Maths Institute (whom you should really follow on Twitter. They perfectly combine stats and smartarse comments), I set about making my own AFL win chance predictor. It’s simplistic (and contains hilarious bugs), but it’s good.
I’m using the mighty Western Bulldogs as my illustrative team, including their game against the
Essendope Drug Cheats Essendon Bombers.
The formula for working out the winning % is…
Win %=(your Win Pts)/(your Win Pts)+(opponent’s Win Pts)
Which means nothing, so let’s work through it.
L: Ladder Difference
A team in 6th is more likely to beat the team in 16th than the team in 11th. So, we take each team’s ladder position (6th for the Dogs, 14th for the Bombers), and subtract that from 19. That way, 1st gets 18 points, 2nd gets 17, and so on.
We then work out the difference, and divide it by 2. That way, the difference is spread evenly across both teams and it’s easier to write an Excel spreadsheet to work it out for you.
S is the streak. If you’ve won three games in a row, you get an extra point. 4, you get 2. Same thing with losing streaks.
H: Home ground
If you’re at home, and it’s not also the opposition’s home ground, you get a point.
F: Fudge factor
We need this to make sure there’s still some level of chance for each team. I set it at 10, which enables it to change enough to make a difference. However, in the case of things like Bulldogs v Brisbane in R23, this led to Brisbane’s win% being negative.
So, there you have it. I’ll use this later in the season for more things, including a simulation of the rest of the year. But, enjoy.