Match Thread [FT] Rangers 4 (Lammers 10'; Danilo 78'; Sima 84'; Dowell 90') - 0 Livingston

[weebluedevil 1,177]

So I think it's fair to say that Rangers XG should be well above the average if the stats include all professional leagues (assuming that includes the lower leagues of some countries). So if 0.41 for the Lammers 'sitter' chance then it probably ranges from 0.1 for the lower league teams to 0.8 for say a Champions league team. I guess the average is in the middle. Rangers therefore should be closer to the 0.8, which is the same as a penalty - fair to say he should have scored. I've ran that clip back a few times (from highlights I could find online) and can only see it from the one angle, and I agree it looks trickier, just because (I feel) Raskins pass is a tad slow, and I think Lammers had to rush the shot due to the back-tracking defender. So if those variables are taken into account, perhaps the XG should be 0.6/0.7, for Rangers, but yes he should score and I would still categorise that chance as a sitter. 

[yuddie 1,025]

21 hours ago, Rousseau said:

'xG' is specifically for football as far as I know. 

 

Probability models could be replicated anywhere, though. I could be wrong. 

Aye I play snooker as a hobby and to keep my head space in check and would love to come up with an 'xP' (expected pot) model for snooker. I feel like analytics could be huge in cue sports haha.

[Rousseau 9,487]

41 minutes ago, weebluedevil said:

So I think it's fair to say that Rangers XG should be well above the average if the stats include all professional leagues (assuming that includes the lower leagues of some countries). So if 0.41 for the Lammers 'sitter' chance then it probably ranges from 0.1 for the lower league teams to 0.8 for say a Champions league team. I guess the average is in the middle. Rangers therefore should be closer to the 0.8, which is the same as a penalty - fair to say he should have scored. I've ran that clip back a few times (from highlights I could find online) and can only see it from the one angle, and I agree it looks trickier, just because (I feel) Raskins pass is a tad slow, and I think Lammers had to rush the shot due to the back-tracking defender. So if those variables are taken into account, perhaps the XG should be 0.6/0.7, for Rangers, but yes he should score and I would still categorise that chance as a sitter. 

This is mistaken. 

 

You can't just pick and chose. This makes a mockery of the stats: 'this lad took a shot, but it's in the Champions League so it's actually worth three shots...'

 

I think the data analysts are in a better position than us to judge the value of the data and metrics.  

[weebluedevil 1,177]

33 minutes ago, Rousseau said:

This is mistaken. 

 

You can't just pick and chose. This makes a mockery of the stats: 'this lad took a shot, but it's in the Champions League so it's actually worth three shots...'

 

I think the data analysts are in a better position than us to judge the value of the data and metrics.  

What I'm trying to say (and I think others are also challenging) is that the 18yr old CB from Elgin, in that position would likely miss that chance. Haaland would very likely score that chance (95% of the time). The stats includes the Haalands etc. and the Elgins etc. and everything in between to create an average. We can decide who would be more likely to overachieve the XG, surely?  Or put another way - what is the XG for Haaland in that situation and what is the XG of a lad from Elgin in that situation, they would be different. 

 

Do you see the stats for the kickers in Rugby?  There's a predictive stat they show on the screen on the probability of the kicker successfully converting the kick. The stat is specific to that player e.g. Russel : 83% chance of successfully converting a kick 40m out, 40 degree angle and 10mph wind speed. That's based on his own kicking data. 

 

What I'm therefore trying to estimate is what the XG is for Lammers in that situation. I expect it to be higher than 0.41. Yes, it's an estimate, not based on proof/data, but on experience of a football follower. If that data did exist (I agree small sample size) it would be higher than 0.4. 

 

Does that make sense? 

 

 

 

 

Edited August 15, 2023 by weebluedevil

[Rousseau 9,487]

6 minutes ago, weebluedevil said:

What I'm trying to say (and I think others are also challenging) is that the 18yr old CB from Elgin, in that position would likely miss that chance. Haaland would very likely score that chance (95% of the time). The stats includes the Haalands etc. and the Elgins etc. and everything in between to create an average. We can decide who would be more likely to overachieve the XG, surely?  Or put another way - what is the XG for Haaland in that situation and what is the XG of a lad from Elgin in that situation, they would be different. 

 

Do you see the stats for the kickers in Rugby?  There's a predictive stat they show on the screen on the probability of the kicker successfully converting the kick. The stat is specific to that player e.g. Russel : 83% chance of successfully converting a kick 40m out, 40 degree angle and 10mph wind speed. That's based on his own kicking data. 

 

What I'm therefore trying to estimate is what the XG is for Lammers in that situation. I expect it to be higher than 0.41. Yes, it's an estimate, not based on proof/data, but on experience of a football follower. If that data did exist (I agree small sample size) it would be higher than 0.4. 

 

Does that make sense? 

 

You're mistaken. You can't apply xG to players. Only the chance. 

 

And then once you start picking and choosing data to fit, it doesn't work. 

[BlackSocksRedTops 3,493]

Can we rename this thread XG? 🤣🤣🤣🤣🤣

[Rousseau 9,487]

1 minute ago, BlackSocksRedTops said:

Can we rename this thread XG? 🤣🤣🤣🤣🤣

What about, 'xG is the best thing since sliced bread'?

[BlackSocksRedTops 3,493]

Just now, Rousseau said:

What about:

 

'xG is the best thing since sliced bread.'

🤣🤣🤣🤣🤣

[weebluedevil 1,177]

1 minute ago, Rousseau said:

You're mistaken. You can't apply xG to players. Only the chance. 

 

And then once you start picking and choosing data to fit, it doesn't work. 

So I think we agree. Saying Haaland had a 40% chance of scoring that chance is incorrect, right? That's because the data is not based on one players data, it's the chance based on many thousands of different players who have missed and scored a similar opportunity to that. So the stats maybe quite accurate for a level in the middle, we will lever know if that equates to the Scottish League 1 or 4th tier in Austria, but should not be accurate when assessing a chance falling to a forward in a Champions League Q game. 

[Rousseau 9,487]

1 minute ago, weebluedevil said:

So I think we agree. Saying Haaland had a 40% chance of scoring that chance is incorrect, right? That's because the data is not based on one players data, it's the chance based on many thousands of different players who have missed and scored a similar opportunity to that. So the stats maybe quite accurate for a level in the middle, we will lever know if that equates to the Scottish League 1 or 4th tier in Austria, but should not be accurate when assessing a chance falling to a forward in a Champions League Q game. 

I still think you're mistaken. 

 

They're only accurate if you include as many chances as possible. 

 

We can't pick and choose. It renders the metric meaningless.