EPL Prediction Model: Wk 30 | Liverpool, Chelsea to win, Arsenal to lose!

EPL Prediction Model: Wk 30 | Liverpool, Chelsea to win, Arsenal to lose!

We’re three quarters of the way through the season and it’s time for a look back. I’ve posted updates each week giving some detail about how the previous round went, but haven’t gone right back to the first week of November and the prediction model’s first outing.

Premier League Predictor Results

Let’s start with the good news. We’re winning! If you’ve bet £10 on every result the model’s called since 1st November, you’re now £140.56 richer.

Here’s what that looked like, from November up to last weekend.

Profit

The model’s done particularly well since a poor few weeks at the beginning of December, which had dragged overall performance into a loss. Partly this is because I’ve improved the way our simulated players pass to each other and how often they shoot, and partly it’s because we build up more data as the season progresses and the model learns with every additional game. Since Christmas Eve – with the latest version of the model in place – we’re up £204 (25%) and have predicted 57% of results correctly.

I’m also pleased to have escaped from the transfer window relatively unscathed. It’s a tricky period, when the model is asked to cope with an influx of new players who it knows very little about and teams’ starting line ups are more difficult to get right.

If all of that has convinced you to take the plunge this week, you’ll be needing some predictions…

15-3

And here are the simulated passing and shooting stats:

15-3 detailed

Finally, the model’s predicted results this week:

Hull City v Manchester City – Away win
Everton v Cardiff City – Home win
Fulham v Newcastle United – Home win
Southampton v Norwich City – Home win
Stoke City v West Ham United – Draw
Sunderland v Crystal Palace – Draw
Swansea City v West Bromwich Albion – Home win
Aston Villa v Chelsea – Away win
Manchester United v Liverpool – Away win
Tottenham Hotspur v Arsenal – Home win