I knew that xG considered the part of the body that you took the shot with, I just forgot to include that.
However, I stand corrected about the position of the keeper and other players. I wasn’t aware that there were xG models that accounted for that - that is very interesting. Though, in my defence, Statsbomb does claim to be pretty much unique in this regard, so it would seem most xG models work the way I described them.
This is a far more useful model, however it does raise a question; when we quote an xG score - what model are we using?
This is in addition to the other main point I mentioned about xG only accounting for registered shots taken. They don’t represent things such as:
- An attacker who slips when through on goal
- A cutback missed by millimetres
- An unnecessary offside from a mistimed run
- A poor final ball to a forward with the goal gaping
All clearly reflective of the creativity of attacking team, but not accounted for by xG.
Of course, over time, a team that created a large number of chances that are not covered by xG would likely
also create a large number of chances that the metric does represent. So in the long term, xG is an incredibly useful statistic.
However, in an individual match, it is possible that xG doesn’t always match “the eye test” because of some of its inherent limitations.