Brain Traysers A while back (page 9 to be precise) I posted a chart showing that under Clement we looked to have tightened up at the back, and at least narrowed the gap between the amount of xG we concede and the amount we generate per game. It wasn't an amazing turnaround, but at least offered some promising signs.
But in the last few games it has gone very very badly. The 3.8 xG conceded against Swansea was the highest amount we have conceded in a game covered by that source (fivethirtyeight). That 3.8 xG conceded is 98.5th percentile bad (away teams only, 9000 observations), with the xGD of -2.8 goals only a smidge better at 97.7th percentile bad for an away team.
As a result, our 6 game rolling xG conceded is now higher than it was when Stam was sacked, and is now above 2 goals per game, higher than it was at any point last season. On the other side, our 'xG for' has been relatively stable at around a goal a game, meaning our xG difference is now -1.26 goals per game. Our rolling xG difference has only been worse once over this period, after Clements 4th game in charge (1-0 away loss at Fulham).
Half a season into a new tenure, and things look statistically worse than when he took over. The only saving grace is that those 6 games are against some in form teams (QPR and Birmingham), and only 2 home games.
I'm not 100% he has to go yet, but after checking those numbers, I'm 98.5%.
while I have my doubts over xG, thank you for presenting the data in such a clear manner. It kind of backs up my instincts - Stam improved things, briefly, before the wheels came off, and Clement has rinsed and repeated.
