Elo Ratings · Win Chances · Spreadsheets for Download
The stats this blog cut its teeth on, the MBBL Elo Ratings are now entering their fourth season. Originally developed in 2016, these use the Elo ratings system developed by Arpad Elo in the 1950s, coming up with ratings and predictions for the T20 Big Bash League. It’s easily our most noteworthy statistic, so we put a lot of effort into it, and here we are ready for another year.
What is the Elo ratings system?
The point of a ratings system is that it takes into account more than just the number of wins you have. That can be overly simple, and doesn’t quite tell the whole picture. Who have you beaten? How’s your form?
The Elo system manages to beat some of these pitfalls. You win points for winning a game, and lose points for losing one. But – and this is its main point, and charm – you get more points for upset wins, and fewer points for expected wins. Also, Elo has a particularly special piece of magic: you can use it to predict win chances. This isn’t just based around an adaptation of it, it’s literally built into the system.
The Elo system was originally designed for chess, where competitors will play many more games than they do in team sports. Based on the calculations Arpad Elo baked into it, a player with a rating 100 points higher than their opponent should win roughly 64% of the time. Narrowing this down to one game, the player would have a 64% chance of winning each game.
These predictions also affect your rating as the games go on. For a game you have an x% chance of winning, you’ll earn (1-x)% of the K factor (how fast/slow the ratings adjust) if you win. Repeat that over the course of many games (eight seasons’ worth, in this case) and you have some ratings.
As can be seen by the complete MBBL history captured above, this season is uniquely placed as the closest tipped season in history. With the Perth Scorchers, the team that’d historically dominated the MBBL, falling apart over the course of MBBL|08, other teams were able to catch up and bunch the ratings together. The Melbourne Renegades lead the competition with 1528, the lowest high score since MBBL|03. At the other end, the Sydney Thunder have a rating of 1483, the highest low score ever.
Unsurprisingly, there are no big margins tipped yet, and we’re fencesitting for pretty much every game. The widest disparity in predictions is the 57% chance the Renegades would have of beating the Thunder, and even then that’s not very strong.
And here’s our prediction for the first game of the season, a clash between two of last season’s underachievers that should still be pretty close:
The Flag 