In the end, it’s kicking a ball. Stats may be useful as an indicator, but it depends how they are calculated. xA and xG seem to have a significant level of subjectivity and complexity.
It reminds me of economists forecasting the economy with tremendously complex models. They always getting it wrong.
The statisticians will spend ages tweaking models to make them better, but it will probably all be rubbish because so much depends on how the manager sets the team up, who the opposition is, were they carrying an injury, had they had an argument with the missus that morning etc.
We know who the good players are. Statistics are always playing catch up and probably only confirm what we know.
