Opening Round Predictions
The opening round of the AFL 2024 season is upon us, with four games ahead all in New South Wales and Queensland. The round features some fascinating match-ups:
Two preliminary final rematches from 2023, Brisbane v Carlton and GWS v Collingwood. Can Calrton or GWS get some revenge for missing the GF last year?
Damien Hardwick’s first foray as coach of the Suns, versus his old team - how long will it take for his system to click?
Melbourne playing their first game of the year after an off season to forget, with questions remaining about their internal culture.
This round also features for the very first time the Informed Stats Early Week Predictions, and the Informed Stats Player Influenced Predictions.
Our early week predictions are based on a teams recent tracked power performance and applying a home ground advantage factor based on historical stats, which provides a relatively solid prediction that aligns nicely with the other predictive models (e.g. Squiggle). Our early week predictions for the Opening Round are:
Swans to beat Melbourne by 7 points.
Brisbane to beat Carlton by 15 points.
Gold Coast to beat Richmond by 6 points.
GWS to beat Collingwood by 5 points.
However, these predictions don’t account for player selections, which could have significant impacts on a teams performance. The most recent example of this is the 2023 West Coast Eagles, a team which suffered from significant injuries, forcing player selections that need to be accounted for to draw down their predicted performance.
Within the Opening Round our early week prediction model is predicting the Swans to beat Melbourne, however the Swans are likely to be without Callum Mills, Luke Parker and Taylor Adams - their captain, a veteran and an experienced off-season recruit. Melbourne on the other hand are likely to play Clayton Oliver, and while the jury is out on how he’ll perform after his off-season troubles, historically he is an outstanding performer, and ranks 3rd in our top 50 players. How will these types of weekly player selection variations be accounted for?
This is where the Informed Stats ERLU model will come in to play. Prior to each weeks games and once the teams have been announced, our early week predictions will be influenced by the named players for each team, and predictions updated to reflect the teams selected.
The ERLU model is based on historical individual player statistics. It is calculated through a scoring system generated by the team at Informed Stats, with the key output being a historical player contribution score. This score is comparable within each team and across the AFL competition, and is reflective of their individual influence to a teams performance. These scores are then used to determine a factor to apply to the predicted outcomes for the week.
At Informed Stats we know that there is too much randomness within the AFL to predict outcomes with extremely high levels of accuracy. The randomness in the AFL is what makes the sport endlessly entertaining. Our aim is to use publicly available statistics to see if we can predict outcomes and player rankings better than others.