United face Tottenham in top-five clash, while City, Chelsea, and Arsenal target favorable matchups. Bottom two West Ham and Liverpool meet in crucial relegation battle.
The xG differentials really tell the story here better than the standings do. That Man Utd vs Spurs analysis where theyre separated by 1 point but the underlying process shows an 8.3 xG gap is exactly the kind of insight that helps separate signal from noise. I remember working on a model for a mid-table club and we had the same issue where results masked serious structural issues that only showed up once you dug into shot quality and territroy control. Good stuff on highlighting the West Ham-Liverpool coinflip too.
Thanks for the read – and I completely agree. xG differential is often the quickest “process vs results” sanity check. I can relate to the issue too: people get carried away when GF/GA is going in their favour because it’s easy to explain and relatable, but over small samples it can be noisy (finishing streaks, keeper variance, game state).
xGD is more about what’s repeatable – chance creation and chance prevention. I remember flagging Man United around Gameweek 6/7 based on the xGA they were conceding even though the GA wasn’t reflecting it. Fast-forward to Gameweek 11 and it looks like that has started to catch up with them.
The xG differentials really tell the story here better than the standings do. That Man Utd vs Spurs analysis where theyre separated by 1 point but the underlying process shows an 8.3 xG gap is exactly the kind of insight that helps separate signal from noise. I remember working on a model for a mid-table club and we had the same issue where results masked serious structural issues that only showed up once you dug into shot quality and territroy control. Good stuff on highlighting the West Ham-Liverpool coinflip too.
Thanks for the read – and I completely agree. xG differential is often the quickest “process vs results” sanity check. I can relate to the issue too: people get carried away when GF/GA is going in their favour because it’s easy to explain and relatable, but over small samples it can be noisy (finishing streaks, keeper variance, game state).
xGD is more about what’s repeatable – chance creation and chance prevention. I remember flagging Man United around Gameweek 6/7 based on the xGA they were conceding even though the GA wasn’t reflecting it. Fast-forward to Gameweek 11 and it looks like that has started to catch up with them.