2 Comments
User's avatar
John Knight's avatar

Great work again, Martin! I’m intrigued by the 0.95 goals per xG. It should be much closer to 1 over a large sample, so I wonder if there is a simple explanation for this?

Expand full comment
Python Football Review's avatar

That’s exactly what I expected as well. I redid the analysis with data from fbref (on 8 seasons instead of 11) and I get 0.97. I mean it’s not shocking. Could be just noise.

Expand full comment