The Elo match predictor generates graphics which show the percentage chance of different outcomes in every English football league match.
It’s designed to provide socially shareable graphics that can be used by journalists at any of Reach PLC’s regional and national titles.
With just one click, journalists you can download either a) a graphic showing the most likely outcomes for ALL matches in a particular division, or b) a Twitter-optimised graphic showing the most likely outcomes for any specific match.
What makes this project innovative?
The poject makes complex mathematics easily accessible to readers through the use of socially shareable graphics. The project acts as a resouce forjournalists to get the graphics for use on social platforms or in articles. It stands alone on a Data Unit server but is't designed to be viewed by readers directly.
What was the impact of your project? How did you measure it?
The graphics have been used by Reach journalists as part of their coverage of local teams.
Source and methodology
The system is fully automated and uses Opta's API feeds to get fixtures, results and league tables. To make the predictions, the gadget uses a Elo-based rating system. This is a system which essentially looks at previous results over a long period of time to calculate a team’s strength. Wins against much higher rated teams result in bigger Elo “gains”. Home advantage is accounted for using historic distribution of home wins, draws and losses. The difference in two teams' Elo ratings is then used to calculate the likelihood of diffrent outcomes by using a logit regression model. Professor Lars Magnus Hvattum provided guidance on the mathematics beind the predictions.
David Dubas-Fisher Cullen Willis Kelly Leung David Ottewell