Better to be Good or Lucky?
If like me you were rubbish at maths and are not a statistician, then this analysis game is often about pinching great ideas from others.
My rule is always to acknowledge the pinchee!
So, my thanks go to @MarcusBr22 and this tweet for introducing me to the concept of “Good / Unlucky” in terms of performance versus Expected Goals.
A reminder that Expected Goals can prove a decent predictor of future performance, and more so than actuals goals for and against, as it focusses on performance (the quality of the chances created and conceded) rather than outcome. We’ll all seen one team dominate to lose to a 35-yard screamer. This is a high risk in a low scoring sport.
By plotting xG For and Against for SPFL teams we can assess whether the outcomes are reflective of the quality of the chances created and conceded.
The plot has 4 corners:
Bad and Unlucky – not only are you creating low quality chances and conceding high quality chances, you are not catching the breaks either. That is, you are scoring less than expected and conceding more than expected.
Bad and Lucky – you underlying performances are poor but you are scoring more and conceding less than expected.
Good and Lucky – You are producing high quality chances and not conceding likewise, and more shot are going in than expected and less being conceded.
Good and Unlucky – your performances are good but you’re not getting the rub of the green.
xG Plot SPFL 21/22
Translated to the SPFL for the season to date:
Firstly, Celtic are quite firmly in the Good / Unlucky category. Having by far the highest attacking xG and the lowest defensive xG, Celtic are not reaping the rewards just yet. That disallowed goal versus Hearts wouldn’t help nor would the penalty miss against Livingston.
That partly explains also the current 4-point gap to The Rangers who are not quite recreating their outrageous xG overperformance of last season. Last term in front of no fans, they scored 92 goals off 68.8 xG – astonishing! Losing the manager and all the backroom staff shrieks “negative variance” so perhaps an adjustment is nigh. We can hope.
Hearts are overperforming and Hibernian underperforming which is consistent with other stats I collect for the league.
Dundee and Ross County look like battling the relegation spot and rightly so based on performance and bad luck – a toxic combination.
These trends can continue for some time, so don’t expect perfect mean reversion.
For Celtic fans, keep the patience.
The underlying data shows great progress under Postecoglou.