Xg from WBA game
- Thread starter anth25
- Start date
Laughing
Well-known member
Loads of things impact the real chance of scoring a goal from any strike on goal. xG only considers some of them.
I don't think anyone would dispute that. Over 46 games and hundreds of shots, it does give you a pretty accurate picture of how a team played in a particular season.
It is also quite good at spotting that a player outscores his xG so is above average. Harry Kane, for example has only failed to beat his xG in 1 season.
This is one of the only reasons I use it for. Also for seeing if a player is getting chances.
For example, you'll see a lot of people saying Lath is feeding on scraps and he's not really getting chances. Yet the xG doesn't really back that up. He's got a total xG now of 3.0 or NpXg of 2.2 so he is getting some chances.
Azaz also got an xG of 1.9 so it suggests he's getting chances as well. If those players had 4 goals between them like their xG suggests they should have had we'd probably be around 6 points better off.
viv_andersons_nana
Well-known member
“He's got a total xG now of 3.0 or NpXg of 2.2” - YOU’LL NEVER SING THAT.
Andy_W
Well-known member
0.83 of it was the Maja chance, xG basically says if you get something on that then it's a goal 4 out of 5 times, which it was basically, the keeper was nowhere. It's difficult to imagine getting a boot on that and it not going in to be honest. It's difficult for xG to know whether he could have got more on that though (as it don't think it factors for this), and had he got nothing on it then it would have been 0 xG. You could evaluate that different, but maybe not also.
It's like a lot of the chances we had were quite good chances, but no massive chances as such. They actually had 3 big chances to our two. Our finishing was actually quite good last night I thought, but their keeper made some great saves. Things like Azaz and Latte Laths header in the first half, neither of those could have done any better, just good saves and unlucky with rebounds.
McGree's was a good chance but there were a couple of players in front of him and the keeper I think, it was no better than Wallace's imo. Can understand why McGree went for a corner and end up missing it.
It's like a lot of the chances we had were quite good chances, but no massive chances as such. They actually had 3 big chances to our two. Our finishing was actually quite good last night I thought, but their keeper made some great saves. Things like Azaz and Latte Laths header in the first half, neither of those could have done any better, just good saves and unlucky with rebounds.
McGree's was a good chance but there were a couple of players in front of him and the keeper I think, it was no better than Wallace's imo. Can understand why McGree went for a corner and end up missing it.
Last edited:
The Maja chance is the perfect example of why xG doesn't work, because the metrics used to calculate the score don't include any "human metrics"...
Distance from goal - He was very close.
Angle to goal - He was pretty central.
Type of assist - The cross was very good.
Body part used to take the shot - Foot (the best body part for that shot).
Whether or not it was considered a "big chance" - It was.
Defensive pressure on shot - There was very little pressure because everyone (including Maja - which isn't taken into account) was too far away.
So on paper, it's the perfect chance which is why it had such a high xG.
It doesn't take into account the distance he was from the ball (too far away), or the contact he was able to make on the ball (minimal), but those type of metrics would be almost impossible to add in.
Distance from goal - He was very close.
Angle to goal - He was pretty central.
Type of assist - The cross was very good.
Body part used to take the shot - Foot (the best body part for that shot).
Whether or not it was considered a "big chance" - It was.
Defensive pressure on shot - There was very little pressure because everyone (including Maja - which isn't taken into account) was too far away.
So on paper, it's the perfect chance which is why it had such a high xG.
It doesn't take into account the distance he was from the ball (too far away), or the contact he was able to make on the ball (minimal), but those type of metrics would be almost impossible to add in.
Colgates_shaving_foam
Well-known member
They only produce data that there is a commercial demand for though so indirectly youtubers and podcasters are definitely to blame. Bloody youtubers.
Laughing
Well-known member
Sports traders, I suspect are the biggest consumers.
Laughing
Well-known member
Is a good example of 1 incident and 1 incident says nothing about the stat more generally.
We had a lot of chances last night but very low %age chances. xG does its job because most of them we were unlikely to score from. Lath's header is a good example because it nearly went in but from where he was and where the ball was crossed from the chance of it actually going in is low. The outcome makes it look like a better chance than it was. We got in a lot of dangerous positions last night but we didn't create many chances where you'd say "should have scored". The McGree one is the only one really. We had a few where we should have done better. A few Lath headers and one for Burgzorg on the ground late on but they are more "should hit the target" than "should score".
We're getting fewer goals than our xG because our attackers are not technically very good. Not only are they not hitting a specific target within the goal like the bottom corner, they aren't even hitting the goal. It's the one area we really need to be more composed and start getting shots on target. We're getting shots from distance on target but usually not with enough power to trouble keepers (Hackney's two goals aside). It's our shooting from in the box which has been erratic.
We're getting fewer goals than our xG because our attackers are not technically very good. Not only are they not hitting a specific target within the goal like the bottom corner, they aren't even hitting the goal. It's the one area we really need to be more composed and start getting shots on target. We're getting shots from distance on target but usually not with enough power to trouble keepers (Hackney's two goals aside). It's our shooting from in the box which has been erratic.
BoroMart
Well-known member
I think there are some data collection errors in that. The idea that 1 in 15 free headers from 12 yards (lath) would result in a goal seems rediculous
Or 3 in 10 for McGrees chance from what 9 yards with the keeper on the floor?
Majas was never a 9 out of 10 score either as the ball was in that awkward too low for a header too high for an easy volley zone. Plus the ball was payed too hard to give the time to get his foot high for a good contact.
xG isn’t perfect over the course of a season these these days anomalies will likely level out, but it’s pretty clear last nights chances break the model
The problem with xG is the stat doesn't change with the keeper on the floor vs. the keeper in the goal with two CBs either side of him?
Frozen Horse
Well-known member
What is meant by assist type?
Does it cover the difference between first time finishes and those when those where the ball is under control, for example?
The infamous Ronnie Rosenthal miss remains the worst I've seen imho, as he had time to control the ball once past the keeper.
That, to me, makes it worse than any number of first time efforts when the ball is flashing across the 6 yard box, even if they closer in with no one to beat.
BoroMart
Well-known member
I believe there are a set of variables that are applied to the location the ball is struck. Defenders between ball and goal, what kind of contact header/shot, vicinity of closest player.
But when trying to replicate a real world scenario in an equation, it will always have finite levels of fidelity and accuracy.
As always it is far more accurate than shots on target, but it is not absolutely accurate and like any algorithm it’s crap in crap out. If the original data has any inaccuracy in location of shot then that will occur in the final xG, and people make mistakes in data entry all the time. You can miniseries that error, but never eradicate it.
XG is a good guiding tool but needs appropriate interpretation.
This was a close game of few big chances, mana missed a booing opportunity but in reality he couldn’t do more than he did which was get a toe on a ball at an awkward height and watch it go well wide. If he hadn’t got a toe on it it still would have gone wide and the xG for wba would be about 0.6. It wasn’t like mana did flunked the chance or the keeper made a great save. In reality it was nearly a scoring chance, McGree the ball sat there and asked to be spanked into the net he flubed a great chance. None of those details are really represented in the xG
Andy_W
Well-known member
The problem you have there is the people who you want to understand that, won't understand that, or they won't acknowledge the reality of it. It's just variance, more tests and more results = more accurate.
As you know, xG is about the best we have which is publicly available and easy enough for most to understand, but we all acknowledge its limitations.
A lot of the people whinging about xG, will have made comments and will go on to make comments about games based on stats like possession, shots, shots on target etc. Obv those are crap by comparison to xG.
Andy_W
Well-known member
Yup.
The thing is, if Maja hadn't even got a touch on it, then people would have thought it was still a good chance, but you get 0 xG for that, as xG wouldn't recognise a cross with no "shot" after it, I think that's how it works.
Then if you ask them if Maja had got a touch (any touch) would you expect it to be a goal, and by what probability, then most would say yes, and I bet it's not far off scoring 80% of the time.
Great cross by Mowatt too mind, thought he was their best player.
It was pretty much an XG of 3, he should have buried it!
Juninhoburger
Well-known member
xG is a daft stat for nerds
Andy_W
Well-known member
I thought it was a worse miss at the time, but looking back it's not that simple.
3 players in the way, one of them the keeper and it's on the half volley on the way up, from a ball coming directly to him, those are never easy.
He had to like thread that in-between/ over the guy on his left, but to the left of the keeper as he looks at him, but needs to lift it too, realistically he only has about 1/8th of the goal to aim at I would say.
It's a 50/50 at best I think, but wouldn't disagree with the 0.33 that xG has it, most of those get blocked rather than miss the target mind. I would rather it had been blocked though, he should still hit the target tbh, for his ability on his good foot.
