I played about in python to scrape game-by-game stats from understat and build some basic formulae to work out how much each team was overperforming or underperforming xG on a 3 and 5 game basis (e.g. over the past 5 games Leeds have, on average, a
goal difference of 1 goal higher
per game than the xG and xGA numbers would suggest).
For the nerdier amongst us, I used Chat GPT to help build the formulae which cycle through the fixtures & compute the rolling 3 and 5 game averages.
The tables are sorted based on their xG over/under performance in the last 3 games
- The higher a team is on these rankings: the worse their goal difference is relative to xG/xGA
- The lower a team is on these rankings: the better their goal difference is relative to xG/xGA
Obviously there's no info here that the betting market hasn't already priced in, but it might put some teams on your radar to consider backing or opposing if you think they might return to the norm in terms of being efficient / inefficient at each end .