The new passing algorithm’s off to a flying start and gave us a late Christmas present on Boxing Day, with six results out of ten called correctly, including Crystal Palace’s away win at Villa.
As is (almost) always the case, the model missed one result by quite a distance, this time in its suggestion that Cardiff would beat Southampton. Managerial sub-plots and teams that don’t turn up on the day are always going to be a problem for a model that uses data on past player performance, but even so that one doesn’t feel like a great call. Still, if you were betting on the predictions in my last post, then you finished +25% and that will do very nicely.
Just a quick post today with predictions for Saturday 28th and Sunday 29th. I’m starting to feel the pressures of fixture congestion! We’ll aim for a wrap up of performance so far in the New Year and we’re also moving into the stage of the season where I’d expect the model to have enough data to work properly – it should be much less volatile throughout the second half of the season.
Here are the percentages for Saturday and Sunday:
And simulated possession and shooting stats:
If you’re betting using the model’s calls:
West Ham United v West Bromwich Albion – Really close. Away win
Aston Villa v Swansea City – Draw
Hull City – Fulham – Draw
Manchester City v Crystal Palace – Home win
Norwich City v Manchester United – Away win
Cardiff City v Sunderland – Draw
Everton v Southampton – Home win
Newcastle United v Arsenal – Away win
Chelsea v Liverpool – Away win
Tottenham Hotspur v Stoke City – Home win
A few draws on the list this week, using our standard rule of betting on the draw when the chances go over 25%
And one big call… Jose to lose his unbeaten home record to Liverpool? The model thinks yes.
New Years Day fixtures will be up once the stats have refreshed for this weekend’s games.