01 - xG model creation

I am able to create an xG model based on your own data. This model could be focus on specific leagues and seasons and it could considers both eventing and tracking data in order to add information about the position of teammates and opponents (defenders + goalkeeper).

The evaluation of the model would consider some shot level metrics like RMSE, AUC and Normalized Brier Score for calibration purpose, in addition to some game and league-season levels as “business metrics”.

Check the following article Fitting your own xG model as a basic example.

It could be also possible to fit some “Possession Value” models like VAEP or xT.