I’m still smiling from the mid-week fixtures. Love or hate the tactics that got West Ham a draw at Chelsea, they were predictable and we saw them coming. I said in my last post that the most likely outcome – by some distance – from the game was a Chelsea win, but that at over 27% probability, the draw was definitely worth a bet.
Overall, we picked six games out of ten across Tuesday and Wednesday, which keeps the average since Christmas at just a fraction over six out of ten results correct every week. One other prediction that I had my doubts about also proved to be pretty accurate. The model gave Everton only a 9% chance in the Merseyside Derby, which felt a little low to me and was about half of where the bookies were pricing them. Although the most likely score was predicted at only two nil, rather than four, Liverpool’s dominance turned out to be a good call.
Ahead of the Liverpool v Everton game, I posted a run from the simulator on Twitter as a bit of fun so people could see some inner workings from the model. This one (just one run of 2000) had a 2-0 result, with Gerrard scoring first (link)
West Ham’s draw marked the single biggest win that the model’s had this season, coming in at 7/1 but since the data started to settle down in the second half of the season and since some improvements to the simulation’s passing and shooting algorithm’s things have been going pretty well.
This is what happened if you bet £10 (it’s a round number) on each of the EPL Model’s call’s in the sixty games that have been played since Christmas.
You’d be sat on a profit of £224.63, or an average of +37% per bet, which is very welcome indeed!
On to this week’s predictions. Here are the percentages:
and simulated passing and shooting stats:
I’ll be backing:
Newcastle United v Sunderland – Fractions away from a draw. Away win
West Ham United v Swansea City – Draw
Cardiff City v Norwich City – Draw
Everton v Aston Villa – Away win
Fulham v Southampton – Home win
Hull City v Tottenham Hotspur – Away win
Stoke City – Manchester United – Away win
West Bromwich Albion – Liverpool – Away win
Arsenal v Crystal Palace – Home win
Manchester City v Chelsea – Home win
A couple of upsets in the list as usual just to keep things interesting. The model’s been consistently overcooking Fulham’s chances all season so please bear that in mind with their win prediction! Some part of Fulham’s poor performances isn’t translating very well into the simulation, possibly because some of their stats this season have been outside the usual norms in the Premier League.
For a reminder of why these bets look the way they do, please see here and I’ll See you next week.