Pearson Out

[ExiledAjax]

16 hours ago, Hxj said:

Or maybe they have significantly better players than average.  Consequently whilst it might be a 0.2 chance for the average player it will a 0.3 chance for one of theirs.  Or maybe with both clubs having lots of goal scorers that works better.

But that is a reasonable conclusion to come to based on the overperformance of the xG data. The degree of over performance does simply show that Arsenal and Burnley are better than they would be expected to be. So yes one reason for that might be that they have better players.

Alternatively, and this applies to your O'Leary example as well, it might suggest that their players are temporarily performing better than the average, and so you can surmise that a regression towards average could be expected.

You can hold disdain for xG, that is fine, and I agree there are limitations to its use and it is often misused. You won't find me saying we "should" have won a particular game based on xG, or that we "will" win a specific future game based on xG. However, anecdotally I can say that since I started looking at in the 19/20 season it has proven to be a pretty solid general long-term predictor of our finishing position in each season. Whenever there has been a general fervent discussion of our impending relegation or play off push I've looked at the xG and shot data and concluded that it's far less likely than people think. In each occasion it's proven true. 

Now that doesn't prove xG to be infallible, it's a tiny sample size and it's biased by my natural inclination to be relatively level-headed. However, for me, I see enough use in xG to believe that it is a) far better use than pure shots/shots on target for/against, and b) is a far better predictor of likely future form than a teams last 8 results would be.

Edited February 7, 2023 by ExiledAjax

[Roger Red Hat]