For love of god please stop the single chance xG analysis

This is really simple. xG models are generally built on logistic regression principles in essence a huge amount of data spits out average outcomes.

With loads of shots the things the model misses is quite insignificant. Over 5 games it's smaller but begins to impact....

In 1 single game the things the model miss are more prominent on the 2 sides xG outcomes hence no models ever really match exactly.

On 1 single shot the variance is ridiculously wild and you've basically taken the statistics principle and shredded it just to fit a narrative