Experimental 361 | Page 2 | Vital Football

Experimental 361

Dreaming of McAuley said:
Flawed - Yes. Nonsense - no. If they were nonsense you wouldn’t almost always see the best teams in the league at the top and the worst at the bottom
Click to expand...

That's the case for many other simpler metrics.

For individual players (See Stockton) they're codswallop.
 
Luke Imp said:
No surprise Freck's top.

Penalties are '1', and Anderson's was a very presentable chance.

Just about mirrors how the game flowed and Crewe's lack of final ball.
Click to expand...
Are you sure penalties are are a 1? My understanding was it was a likelihood out of 1 so a penalty would be like 0.7
 
Dreaming of McAuley said:
Are you sure penalties are are a 1? My understanding was it was a likelihood out of 1 so a penalty would be like 0.7
Click to expand...
Could've sworn I read it was a 1, although having said that, logic would suggest it'd be 0.8 because I can't recall Freck having a shot other than the penalty?
 
yes, flawed, inconsistent, or whatever:
pett was on the pitch for 33+ minutes, scored, and received Xgoal of less than 0.5
stockton was on the pitch for 45 minutes, did nothing, and received Xgoal of 0.4
 
Luke Imp said:
Could've sworn I read it was a 1, although having said that, logic would suggest it'd be 0.8 because I can't recall Freck having a shot other than the penalty?
Click to expand...

It entirely depends on the model being used. There isn't a definitive model to calculate XG, many betting pundits have built their own, I suspect Ben has as well. I don't believe anyone has a penno at 1 though.

So we have a series of models built to predict outcomes of matches that come with all sorts of caveats about what has been included and left out. Should they be taken with a healthy dose of scepticism? Absolutely. Are they completely meaningless for predicting the outcome of a match? Clearly not, professional bettors use them all the time as a tool...
 
Imps2018 said:
Is that saying expected goals would have had it 2-2 with rounding?
Click to expand...

That's one interpretation but I'm not sure rounding applies to XG. You could also say after rounding there was a 1 goal difference. Their 0.8 at the end for the penalty made a big difference.

The thing about our goals is 4 from corners and 1 from a long throw. They won't be worth much in expected goals. We had 13 corners - I bet aggregated that would count for little more than 1 expected goal but we scored 4.
 
Skip155 said:
That's one interpretation but I'm not sure rounding applies to XG. You could also say after rounding there was a 1 goal difference. Their 0.8 at the end for the penalty made a big difference.

The thing about our goals is 4 from corners and 1 from a long throw. They won't be worth much in expected goals. We had 13 corners - I bet aggregated that would count for little more than 1 expected goal but we scored 4.
Click to expand...
Bozzy’s goal from the corner was 0.7 ish surely? Don’t remember him having many other chances.
And we gave away the penalty so can’t use that as an excuse.
To me this says it should have been a 2-2 with Imps grumpy they couldn’t get that 3rd. Whereas anyone who watched it would likely agree we were unlucky not to hit double figures.
 
Imps2018 said:
Bozzy’s goal from the corner was 0.7 ish surely? Don’t remember him having many other chances.
And we gave away the penalty so can’t use that as an excuse.
To me this says it should have been a 2-2 with Imps grumpy they couldn’t get that 3rd. Whereas anyone who watched it would likely agree we were unlucky not to hit double figures.
Click to expand...

You can interpret it how you like - that's data. The Bozzie one is interesting - probably recorded as a shot on the goal line - he won the second ball - not direct.
 
Skip155 said:
You can interpret it how you like - that's data. The Bozzie one is interesting - probably recorded as a shot on the goal line - he won the second ball - not direct.
Click to expand...
Don’t get me wrong I love the use of stats in sports and work in data/stats stuff at the moment. I just feel this model needs a lot of work at the moment.
 
certainly wouldn't want to be using these stats in isolation to predict anything.
whitfield's chances of scoring are only high because they won a penalty.
bostwick's chances of scoring are only high if we win a corner and the ball falls to him unmarked on the goal line.
where is mccartan's figure, the guy that beats a couple of players and fires into the bottom corner from eighteen yards? that goal appears to accumulate next to nothing.
 
nlondonimp said:
certainly wouldn't want to be using these stats in isolation to predict anything.
whitfield's chances of scoring are only high because they won a penalty.
bostwick's chances of scoring are only high if we win a corner and the ball falls to him unmarked on the goal line.
where is mccartan's figure, the guy that beats a couple of players and fires into the bottom corner from eighteen yards? that goal appears to accumulate next to nothing.
Click to expand...
Exactly, and that's the point. McCartan's was a more difficult goal, and as such has a lower XG rating.

Your XG will be higher if you create 4no Boswick chances than 4no McCartan chances because the former were easier finishes that you'd expect to result in goals 99% of the time.

Of course, they'll be the odd anomaly where you dominate and end up scoring when a player pings one into the top bin and you win 1-0 with a low XG but by and large, the better quality of the chances, the more likely you are to win a game and that's what it's showing.
 
I think the problem some people are having is not understanding these models and what they are supposed to show. If we get 13 corners then Bostwick will have a pretty decent XG. Same I imagine for Shackell had he been playing and perhaps Wharton if he continues to score from corners.
 
You must log in or register to reply here.