- Oct 9, 2013
- 10,942
- 14,851
- AFL Club
- Western Bulldogs
Most of what you say is right.I've never placed much weight on expected score. Expected Score is effectively a long sequence of sliding door moments within a match.
As soon as you allow for the possibility that just one of those shots had turned out differently you have to concede that the rest of the match would then have played out differently. That's because what was a kick-in would now be a centre bounce instead. Or even a boundary throw-in.
And vice versa. After Bailey Smith failed to handle the ball cleanly in the frenetic third quarter of the 2021 grand final, if the Melbourne shot on goal a few seconds later had been a behind instead of a goal they might never have gone on their 16 goal surge. Instead we might have taken a safe kick-in, chipped it around the boundary line and ... well, who knows how it might have worked out.
So you can't validly add up the expected score from every shot in the match and then say "we should have won that by 4 points instead of losing by 17" (or whatever).
I'm not into Expected Score ladders either.
However accumulated expected score (the expected vs actual differential) might still be useful as a season-long measure. Over a whole season it's possible it will provide a rough guide to which sides have a problem with accuracy and which ones don't. I expect most sides would regress to the mean though.
It's not right that most sides would regress to the mean though. Like every data set, it will have a distribution around the mean. This will apply over whatever time period you want to set.
It definitely is a useful measure for the purpose you describe (bolded).
When looking vs another team within a game, or season though it's just a rough guide or estimation.