Snowflake Royal sandman Snowflake Royal I'm always rather suspicious about expected goals. It seems a bit too faddy and unable to take account of quite important variables.
It's just so. What's the word... American.
Yes. That too.
Hah - possibly an americanism, but haven't foreign imports been a net benefit to British football, or should we all go back to old school 442 hoofball?
WRT xG, the explanatory power is indeed pretty poor (a readily available model critiqued by
this article had an r-squared of only 0.165), but when compared to any other easily quoted statistic it is way more powerful. Indeed some providers have started to include the missing variables referred to (including GK placement, direction and intensity of defensive pressure on shot, striking foot preference). Indeed, that article has a fair rebuttal,
here.
Possession tells you nothing about where you had the ball and what you did with it (Stamball proved that), number of shots can be easily inflated by teams/players who make terrible decisions (Fulham have had 2 more shots this season than Arsenal), shots on target are marginally better, but the same issue applies.
xG while basic and flawed can tell you a lot of things:
- Mo Salah over-performed last season/was likely to drop off this season (32 2016/17 league goals from 25 xG)
- David De Gea single-handedly got Man U into the CL (28 conceded vs 43.5 xGC, worth 19 points more than their 6th placed xP)
- Burnley were either incredibly good (/world class) defensively - the pundit's since they got promoted back to the PL - or have ridden their luck for an extended period of time (13xG fewer than actually conceded last season), the latter looks more likely this season.
The above ones are obvious/established narratives than can be supported by xG, but it can also be a really helpful tool to tease out the less obvious (or counter wrongly established) ones:
- Morata and Sterling are statistically not bad finishers, in spite of the occasional bad miss (xG and G consistent)
- Jadon Sancho and Reiss Nelson look very promising at a young age playing in a top league, but their numbers suggest its premature to declare them an elite level winger-playmaker and striker (JS has 6 league assists from 2.65xA, RN has 6 league goals from 1.8xG), so expect both to drop off
- (and most importantly) Reading continue to be incredibly lucky / Paul Clement is a tactical genius - you chose the narrative.
WRT the OP, even though he is available and has a very strong CV, no thanks. Until the fans start to turn on PC (as opposed to the current state of apathy), no need to reset - we need stability to rebuild, even if that risks another relegation fight.