What to expect this season

After a bit of a gap year in my blogging activity I am keen to get back to the sort of things I was doing during the 15/16 season – exploring machine learning techniques that might help us identify value bets. Since we have a month before the new season starts, and because my models have evolved considerably last season, I thought I’d go through the basics again.

In the next few weeks I will post a series of articles reviewing how to build a model that predicts the probability of winning, drawing or losing for a given team. My aim this time round is to give enough information (and data) for a keen reader to build a model that competes with mine… so sharpen your R skills and stay tuned!

I will then write a refresher on how such a model can be used to identify potential value in the market and we will go through how to set a staking plan again. Most of this info however is already on the blog and I don’t plan to change much for the coming season.

Finally we will start betting again! I will be posting model probability estimates each week, starting in October (after the model has warmed up) for the 5 major European leagues (England, Spain, Germany, Italy and France). I will also provide these estimates to Alex (@fussbALEXperte) weekly for his model comparison project. Oh and a big thank you for the art work Alex!

 

The first post will be on feature engineering and we will look at what might be good predictors of a game outcome. The idea is to refresh this with the work I have done last year. Hopefully coming very soon…