Bluedylan1 wrote: ↑Wed Nov 29, 2023 12:51 pm
I disagree. I do think we're making consistently good chances.
Sure, we don't make them in an attractive way and they are usually the result of a knock down or a bit of ugly play, but that's kinda a different point.
Nah it’s very much part of the same point.
A shot on goal 8 yards out with no players between you and the goal is a high xG chance.
If that shot comes after a knockdown/scramble and you’re reacting to the new ball trajectory then it’s a lot worse chance than if it’s been squared to you and you have time to set before hitting it, yet they’ll have the same xg
Bluedylan1 wrote: ↑Wed Nov 29, 2023 12:51 pm
I disagree. I do think we're making consistently good chances.
Sure, we don't make them in an attractive way and they are usually the result of a knock down or a bit of ugly play, but that's kinda a different point.
Nah it’s very much part of the same point.
A shot on goal 8 yards out with no players between you and the goal is a high xG chance.
If that shot comes after a knockdown/scramble and you’re reacting to the new ball trajectory then it’s a lot worse chance than if it’s been squared to you and you have time to set before hitting it, yet they’ll have the same xg
I take the point, but I disagree.
We're making clear cut chances. The weekend for example - Dom's headers, Doucoure's shot just wide, McNeil from a few yards out. They are objectively good chances. No, it's not a striker going through on goal against the keeper in oceans of space, but those chances are very hard to create unless you have a certain quality of player.
For our more agricultural ''style'' of play (if you can call it that), a couple of largely unchallenged headers at close range and a couple of close range shots in the box are excellent chances.
4evablu wrote: ↑Tue Nov 28, 2023 2:01 pm
What is the factoring of what is a chance and what is the degree of difficulty of a chance?
AFAIA, it's a data and probability generation model.
If, in X amount of attempts, somebody has shot from that position before then an xG of 1 means it has gone in every time and 0 is a statistical and legal impossibility.
So 0.9 is 9 out of 10 people who have had a crack from that position have scored
The "position" takes the below into account:
Distance to the goal
Angle to the goal
Did the player strike it with his feet or was it a header?
In what passage of play did it happen? (e.g. open play, direct free-kick, corner kick, counter-attack)
Has the player just beaten an opponent?
I've heard there's more since I checked. But I think that's around GK data, I don't actually know.
Thanks mate - when i've looked at it for me it doesn't show the actual definition of a chance (if that makes sense)
eg: From a corner/free kick DCL is nowhere near the goal, he jumps and heads the ball towards goal from a very acute angle marked by say 2 defenders...never even looked like a possible "chance" but he made it an effort towards goal. may or may not be saved - may or may not go wide - is that considered/included as a chance ?
Against the ball crossed into the box clear header on goal - that's a definate chance...
That's why i don't go with this xg stuff.
4evablu wrote: ↑Tue Nov 28, 2023 2:01 pm
What is the factoring of what is a chance and what is the degree of difficulty of a chance?
AFAIA, it's a data and probability generation model.
If, in X amount of attempts, somebody has shot from that position before then an xG of 1 means it has gone in every time and 0 is a statistical and legal impossibility.
So 0.9 is 9 out of 10 people who have had a crack from that position have scored
The "position" takes the below into account:
Distance to the goal
Angle to the goal
Did the player strike it with his feet or was it a header?
In what passage of play did it happen? (e.g. open play, direct free-kick, corner kick, counter-attack)
Has the player just beaten an opponent?
I've heard there's more since I checked. But I think that's around GK data, I don't actually know.
Thanks mate - when i've looked at it for me it doesn't show the actual definition of a chance (if that makes sense)
eg: From a corner/free kick DCL is nowhere near the goal, he jumps and heads the ball towards goal from a very acute angle marked by say 2 defenders...never even looked like a possible "chance" but he made it an effort towards goal. may or may not be saved - may or may not go wide - is that considered/included as a chance ?
Against the ball crossed into the box clear header on goal - that's a definate chance...
That's why i don't go with this xg stuff.
I think it's anything that involves an attempt to score a goal, but doesn't include attempts where the attacker makes no contact with the ball at all, so missing a cross completely and making no contact.
The scoring takes into account the situation you described (the chance that didn't look possible) by giving it a very very low score. So that might get a 0.05, where as the chance Dom had at the weekend where he was 6 yards out and unchallenged, might score a 0.4 or 0.5, as roughly half the time you'd expect someone to score in that situation.
Where it doesn't really help us (if it works the way I think it does) is all the chances that were good opportunities but a real effort on goal didn't happen. So a low cross that skips across the box about 5 yards out, but is missed by everyone who are just too far away to make contact, isn't included at all.
Would it treat these two instances the same, using an example similar to Dom's headed chance, corner ball to the centre of goal six yards out, all else being equal...
a)floated ball, Dom wrestling with 2 opposition players (McGuire & Lindelof), a packed box, moving away from goal he struggles to make sufficient contact, ball dribbles to the keeper
b) a driven ball, Dom largely unmarked and has speed and a clear run from the penalty spot to connect with it at pace
If they're treated the same, or even similarly, then it shows the flaws in the system. Penalties at 0.76 also skew.
Even different models can produce widely different results based upon the assumptions they use in them. It's far from definitive, and like anything of this type of statistic there is a settling over time as the sample size increases.
Which is all true to some extent, but also not hugely important.
It counts shots, some shots are more likely to result in goals than others. It is largely accepted by the best minds of the biggest sport on the planet as *a metric* for tracking attacking output and defensive effectiveness, beyond gf ga, but if you poke it enough, cracks will appear. If those cracks outweigh the overall usefulness of the metric I don't know to be honest I doubt it, but cracks will appear.
I think it's like most statistics they are maneuvered to suit a purpose and a strap line.
Bottom line is we don't score enough goals, end of, to win certain games. It's not feasable to expect to score every "chance" but if we'd of taken 25% of our chances on saturday we'd of won.
The other flaw in xG is it aggregates historical data from all players, meaning it doesn't account for the player's own ability - that's why Haaland will always outperform his xG, while Maupay gonna Maupay.
Trowel wrote: ↑Thu Nov 30, 2023 9:18 pm
The other flaw in xG is it aggregates historical data from all players, meaning it doesn't account for the player's own ability - that's why Haaland will always outperform his xG, while Maupay gonna Maupay.
Enough data does end up accounting for that though.
At a certain level it will average out to being the average player. (That level has already been reached)
What makes Haaland 'better' than Maupay is that he is able to score goals much lower on the xG chart with more regularity.
And this is where we start getting into the reason it's actually useful.
Over a couple of seasons it can confirm if a player is just on a run of form or not. Can also show if a scorer is a natural finisher or just somebody who grafts (or works in a system) to create loads of chances
