Useful article showing the benefits of expected goals and how it can help predict and evaluate performance (pretty spot on about Man Citys weaknesses last summer and Chelsea's regression to the mean, using expected goals as a reliable predictor):
Chelsea lost Costa replaced with someone massively inferior to him, Cahill is a year older, Bakayoko will be adapting and Fabregas is ageing, they didn’t look good to begin with.. don’t need in depth stats to tell you that.
It’s like using stats to review last season’s Arsenal defence at the start and saying they won’t win the league. They won’t win the league because they had a terrible defence, individually they’re not great players and Cech’s reactions are not what they used to be.. again you didn’t need in depth stats to tell you that.
Stats do have some use in the game granted, especially for tactics, but the majority of the time they tell you very little as football is an unpredictable game.