Jan 2020 Premiership football prediction model


Hard to believe it’s actually the year 2020!
It’s a date used historically in sci-fi, and particularly a Cyberpunk RPG game I loved as a teen… so excited for the Cyberpunk 2077 title that’s coming in September!

Games I’ve been playing:

  • Modern Warfare – today reached level 155 in Season 1; heaps of fun, but the updates have been a nightmare having to download DLC packs multiple times
  • Sekiro – when I’m not too scared to play, usually while waiting for Modern Warfare downloads. I’ve explored and progressed a little further, to be humiliated by another boss.

Premiership football prediction model
Continuing my focus of scraping football data, focused on the English Premiership, inspired by this Medium post: Predicting Football Matches using EA Player Ratings and Tensorflow
I’ve collected heaps of data with Python using Beautiful Soup and Scrapy, including:

  • game results back to 1936
  • player rosters for each game
  • team stats
  • player stats
  • & odds

I’ve tested and trained an xgboost model base on Premiership games from 2006 to 2018, then validated the model on 2019/20 season to date.
Currently looking at ~70% accuracy, predicting either a Home win, Away win, or Draw, then built a betting strategy based on these results.

The number look very interesting, and I am to place my first bets next week 🙂
My wife has allowed my $100 to start with, if all goes to plan this time next year I’ll have increased that to beyond $1 million :p

I will upload my scripts to GitHub, although I seem not to be able to currently access my account 🙁