• The first stage of the forum upgrades has now been completed but they remain in a degraded state and are still being worked on. Normal posting/reading should now be possible.
    Please read this thread for more details.
    New user registrations are currently disabled.

This xG nonsense

RossTheNinja said:
I'm trying to give it a chance but warra load of shite. It was basically the same shot he scored the other week.
Click to expand...

xG of 0.01, but 0.84xGOT.

Basically what that means is the chance of scoring from where he had the shot was 1 in 1000, but the actual shot goes in 84 times out of 100.

A very difficult place to score from, but it was an excellent effort.
 

AC Mack said:
I think that both shots would have had a separate xG score, and an overall average would have been attained.
Click to expand...
I don't know how they calculate it.

The way mathematical expectation works for a double event of value 1 (like a goal), where the success of the first part means there can be no expectation from the second, (If Brobbey scores first hit, he doesn't get the second chance, so they cannot be just added) is below.

xG = P(goal first shot) + P(misses first shot) x P(goal second shot). (P is probability - I've used this so as not to confuse the 'x' in xG with a multiplication sign)

If, for example, the xG for the first shot was 0.7, and the second shot was 0.6
then when totalling the xG for the match it should be 0.7 + 0.3 x 0.6 = 0.88

Whether they do this or not, that's the maths.
 
Quillips said:
I don't know how they calculate it.

The way mathematical expectation works for a double event of value 1 (like a goal), where the success of the first part means there can be no expectation from the second, (If Brobbey scores first hit, he doesn't get the second chance, so they cannot be just added) is below.

xG = P(goal first shot) + P(misses first shot) x P(goal second shot). (P is probability - I've used this so as not to confuse the 'x' in xG with a multiplication sign)

If, for example, the xG for the first shot was 0.7, and the second shot was 0.6
then when totalling the xG for the match it should be 0.7 + 0.3 x 0.6 = 0.88

Whether they do this or not, that's the maths.
Click to expand...

They don’t. They just add the xG from all the shots.
 
XG per AI in terms of for and against has us at 34 points after 31 games. So fair play we are more clinical by 9 points stunning really. It’s not overly surprising I guess as we’re second bottom in the h league for goals scored.
 
DonaldDownTheWing said:
They don’t. They just add the xG from all the shots.
Click to expand...
If so then this is a big xG limitation. The 'bobbles in box' where there are 2 or 3 chances to put it away in a few short seconds are being overvalued - not each one individually, but when totalled them for a match. You'd sometimes get the ludicrous situation of being credited with an xG of more than 1 for a single phase of play!
 
Quillips said:
If so then this is a big xG limitation. The 'bobbles in box' where there are 2 or 3 chances to put it away in a few short seconds are being overvalued - not each one individually, but when totalled them for a match. You'd sometimes get the ludicrous situation of being credited with an xG of more than 1 for a single phase of play!
Click to expand...
But they are in fact 2 or 3 chances to score
 
So actually, on FotMob there are:

- 5 shots within the 6 yard box which add up to 2.28 xG.

- 6 shots within the penalty box which add up to 0.55 xG

- 6 shots outside the box which ad up to 0.16 xG

Total = 2.99 xG (cumulative)
However the xG listed is 2.44xG

Looking at the 2 Brobbey chances, they add up to 1.5xG (0.69 + 0.81)

If you follow the correct maths to work out the probability over the 2 shots it would be 0.61 + (0.39 x 0.89) = 0.96xG

1.5xG - 0.96xG = 0.54xG (the difference between the cumulative total of the 2 chances and the probability of the 2 chances)

2.99xG - 0.54xG = 2.45xG

I think it is clear that actually they are using an advanced model to calculate xG, not simply cumulative.
 
King for a Day... said:
So actually, on FotMob there are:

- 5 shots within the 6 yard box which add up to 2.28 xG.

- 6 shots within the penalty box which add up to 0.55 xG

- 6 shots outside the box which ad up to 0.16 xG

Total = 2.99 xG (cumulative)
However the xG listed is 2.44xG

Looking at the 2 Brobbey chances, they add up to 1.5xG (0.69 + 0.81)

If you follow the correct maths to work out the probability over the 2 shots it would be 0.61 + (0.39 x 0.89) = 0.96xG

1.5xG - 0.96xG = 0.54xG (the difference between the cumulative total of the 2 chances and the probability of the 2 chances)

2.99xG - 0.54xG = 2.45xG

I think it is clear that actually they are using an advanced model to calculate xG, not simply cumulative.
Click to expand...
Appreciate xG but that got a bit complicated for me, did we still win?
 
MadMackem7 said:
But they are in fact 2 or 3 chances to score
Click to expand...
Yes, they are. But chances cannot simply be added together to give an overall expectation if the outcome of one chance effects another one. If they are part of the same phase/move then, in maths language, they are not 'independent events'. Whether the first shot goes in effects whether there will even be a second or third shot in that phase. In such cases 'conditional probability' must be used.
 
You must log in or register to reply here.
Back
Top