Moneyball and Footy

[Nano]

Stats can be a magnificent tool. Unfortunately, most people don't understand them. Using data badly can be worse than not using it at all.

People get fixated on specific stats as well, like xG, and extrapolate all other data points from it instead of being aware that the xG itself is a dynamic statistic.

Depends on the stat, if DeChambeau is told from his stat people that he is 1.4% more likely to get a birdie in a Thursday, what use is that?

That is of no use but if the stat was that he can hit the fairway 90% of the time if he drives it 250yds and he can hit it 10% of the time if he drives it 350yds and his chance of getting up and down is higher from the fairway then you'd suggest he should do what he can to hit the fairway but if his chances of getting it up and down from 100yds from the rough is the same as 200yds from the fairway then hitting it 350 into the rough is a better option than 250 in the fairway because the 10% he doesn't hit the fairway at 250 is much harder to get up and down than the 10% he's in the fairway at 350.

 

[Brian Marwood]

That is of no use but if the stat was that he can hit the fairway 90% of the time if he drives it 250yds and he can hit it 10% of the time if he drives it 350yds and his chance of getting up and down is higher from the fairway then you'd suggest he should do what he can to hit the fairway but if his chances of getting it up and down from 100yds from the rough is the same as 200yds from the fairway then hitting it 350 into the rough is a better option than 250 in the fairway because the 10% he doesn't hit the fairway at 250 is much harder to get up and down than the 10% he's in the fairway at 350.

I will only read on average 4.876% of replies that fail to use commas.

 

[Frozen Horse]

I think Moneyball the book, and the original Sabermetrics are often misunderstood.

They didn't invent the use of statistics in player scouting etc : statistics had always been used.

Instead, they started to ask which statistics really mattered? Batting average, commonly used, didn't seem to make much difference while on-base percentage did.

Likewise, I suspect possession doesn't matter per se, possession in the opponent's half might. I don't think you can improve a team's performance simply by improving their possession statistics. You can improve those stats by stringing the ball across your defence for 90 minutes, but you won't win more matches. You might win more matches by improving possession in the opponent's half or the final third, but that's hard to do and, in reality, means improving the quality of your play, not just manipulating statistic. The stats are a description of what is happening, not the cause of them.

As for xG, I'm not that convinced of what its for. As I understand it, it's a way of modelling performance if finishing didn't count. But of course, finishing does count: it's the single most valuable skill in the game. Where I can see it might have some value is in measuring an individual's finishing. If we know a striker's xG, and his goalscoring is significantly above that, maybe we can believe he's an above average finisher.