Expected goals

[Leisure]

Please can someone explain for me just what are Expected Goals and what is the point of the stat?

[Buxtonclaret]

I'll look with interest here too, Pete.

[Devils_Advocate]

Im not 100% sure but I think 10 Opta analysts watch each game and if they think there is a good goalscoring chance they press a button. If 3 press the button for the same chance then that is a 0.3 opportunity or if 7 press the button its 0.7 and so on.

At the end of the game they add up all the scores and the total is the expected goals. So if Burnley created 4 good chances and they were rated 0.3, 0.4, 0.6 and 0.8 our expected goals for the game would be 2.1 goals

[Roosterbooster]

It's a reflection of the quality of the chances created - based on the chances created, how many goals might you expect that to result in

Normally you'd expect xG and actual goals scored to show a correlation, especially over a season

But in specific games, players might miss easy chances, or the keeper might pull off some absolute worldies. Or vice versa. Someone might score a worldie, or the keeper might make a calamitous error - and xG might be miles off

People read too much into it. But it can be a useful tool. I've seen teams lose 4-0 for example. But xG has been quite well matched. And that reflected the game overall, only for one team to score 2 worldies, and the keeper make 2 or 3 unbelievable saves (as an example). Pundits get carried away with scorelines. But on another day, the game could have finished 2-2, or even 3-0 or 4-1 the other way

I've heard pundits say "xG doesn't matter. The only thing that matters is how many goals you actually score". And while that's true, these pundits are missing the point

[Aclaret]

As Leisure suggests......what's the point.

[RVclaret]

Please can someone explain for me just what are Expected Goals and what is the point of the stat?

When a player goes through one on one with a keeper, you’d say that’s a very good chance right? In comparison, a player who takes a first time pot shot from 25 yards with 5 players in front of him, probably isn’t a good chance. What if the first player 1 on 1 misses the target, second effort hits the target, straight at the keeper, and trickles through?

One of those gets recorded as a shot on target. Shots on target aren’t really much use in that case when analysing performance. Expected goals provide a way of measuring chance quality, that’s pretty much it. Statistical models calculate probabilities of scoring, using what’s happened previously by players in the same situation. 0.75xG is roughly what a penalty is as statistically 3 in every 4 pens are scored. 0.0003 (example) xG is awarded to an overhead kick from the edge of the box because not many of them are scored.

Using this, over time, it can be used as a metric to see if a player or team over performs or under performs their xG. The idea being that you may be scoring lots of worldies if you overperform by a massive amount and analysts may suggest ‘variance’ leads to reverting to the mean and that team / player is just ‘running hot’ rather than their current form being a ‘true’ reflection of ability.

Teams like Brentford stopped using actual league tables and created models using xG (in conjunction with other things) when measuring their performance.

[dsr]

For the Luton game, both the first half goals were rated about 0.2, which means that they would have expected Brownhill and Odobert to score about 1 chance in 5 when they are put through like that.

"Rubbish", you might say. And I wouldn't argue.

[ab1882]

https://statsbomb.com/soccer-metrics/ex ... explained/
https://theanalyst.com/2023/08/what-is- ... d-goals-xg

[PremierLeagueClass]

For the Luton game, both the first half goals were rated about 0.2, which means that they would have expected Brownhill and Odobert to score about 1 chance in 5 when they are put through like that.

"Rubbish", you might say. And I wouldn't argue.

I did think our expected goals seemed very low for that game.

[Notsosuperstevedavis]

They supposedly analyised every shot and every chance since the prem started. A shot that goes in x1 in every 100 would be worth 0.01 xG. A pen on the other hand is probably worth 0.95 xG.

They look wt the shots we’ve had and accumulate your xG as a sum.

If you’ve had shots that equate to 1.2xG when added but you’re winning two nil. You could argue that you’ve been lucky so far, their keepers having a mare.


In short. Its a stat for its 0-0 in my eyes. The team with a bigger xG ‘should’ be winning.


Maybe means more when measuring keepers. If they’ve faced shots equal to 5.0xG but it’s actually 1-0. He’s playing out of his skin.

[Leisure]

Im not 100% sure but I think 10 Opta analysts watch each game and if they think there is a good goalscoring chance they press a button. If 3 press the button for the same chance then that is a 0.3 opportunity or if 7 press the button its 0.7 and so on.

At the end of the game they add up all the scores and the total is the expected goals. So if Burnley created 4 good chances and they were rated 0.3, 0.4, 0.6 and 0.8 our expected goals for the game would be 2.1 goals

Just taking this example - how can you have 2.1 goals???

[Goliath]

It's basically an upgrade on the old shots on target statistic which was utterly useless. Xg is a much better indicator in regards to which team had the better chances over 90 minutes. Whether this means the team with the better xg deserved to win is debatable.

[Devils_Advocate]

Just taking this example - how can you have 2.1 goals???

The same way you can have 68% possession

[Leisure]

The same way you can have 68% possession

Not the same at all! Show me a game where a team scored 2.1 goals!

[bfcjg]

File under low block, high press, false number 9,inverted full back playing through the channels etc etc etc bl00dy etc.

[ElectroClaret]

File under low block, high press, false number 9,inverted full back playing through the channels etc etc etc bl00dy etc.

Yeah.
What's the point in over-analysing things in what should
be a very simple game.

[dsr]

Just taking this example - how can you have 2.1 goals???

"Expected" in this context means the same as "average".

[dsr]

They supposedly analyised every shot and every chance since the prem started. A shot that goes in x1 in every 100 would be worth 0.01 xG. A pen on the other hand is probably worth 0.95 xG.

They look wt the shots we’ve had and accumulate your xG as a sum.

If you’ve had shots that equate to 1.2xG when added but you’re winning two nil. You could argue that you’ve been lucky so far, their keepers having a mare.


In short. Its a stat for its 0-0 in my eyes. The team with a bigger xG ‘should’ be winning.


Maybe means more when measuring keepers. If they’ve faced shots equal to 5.0xG but it’s actually 1-0. He’s playing out of his skin.

Not necessarily. They may have had expected goals of 5 but they have missed a load of sitters. There's a different calculation for the number of goals that the keeper would expect to let in.

[IanMcL]

It is what we expect when we say the score, first goal scorer and correct minute!

[Bin Ont Turf]

It's for those sitting at home to feel that they are involved.

[fidelcastro]

Basically it's a load of crap.

[RammyClaret61]

Not the same at all! Show me a game where a team scored 2.1 goals!

Exactly. The score at Luton was 1-4.

Not Luton 1.43 Burnley 1.46.

With all the Americanisation now creeping in. We know they don’t like draws in their sports. Will they use this to decide drawn games? They love their stats.

[daveisaclaret]

If you don't like the stat you can just ignore it instead of pretending not to understand it

[RammyClaret61]

If you don't like the stat you can just ignore it instead of pretending not to understand it

I’m not pretending to not understand it. I genuinely don’t see the point of it, I’ve tried to understand it, but I don’t see what it gives to anyone. How can a stat that says Burnley who score 4 goals in a match were expected to score 1.46?? Like Leisure says, WTF is .46 of a goal?? And would Rob Jones disallow it?

[Bullabill]

You can have a score of 1.46 goals if you have 127% possession.

[Clive 1960]

funny how the game as changed 2.1 goals load of tosh , never other think a game of football it's easy to play....

[SammyBoy]

That year Blackburn were third after 26 games with a minus 4 goal difference their XG was terrible, which confirmed to me they were in a completely false position relative to their performances.

[Mattster]

Always makes me laugh the reaction xG gets from a certain section of football fans.

"How can you score 0.1 of a goal?!?"
"How long until they start using it to decide draws?!?"
"Won't somebody please think of the children?!?"

1200x1200bb.jpg (205.24 KiB) Viewed 2865 times

It's just a statistic, like the number of shots, number of passes, possession, number of corners, number of offsides etc. Just like those statistics they've never taken issue with, in isolation it means little and no one is forcing you to pay aby attention to it. But unlike those statistics it has a decimal point it, therefore it is the product of nerds!

[Leisure]

If you don't like the stat you can just ignore it instead of pretending not to understand it

Who's pretending!

[BabylonClaret]

Sure I saw a comment t last year where so eone had poi Ted out our xg:goals ratio was quite high Iin the Champ (eg. How many of those Benny wi ers were there which would be low xg) and therefor we scored a lot fro. Outside the box which are .uch harder to convert in the top league because you get less space (unless playing against is of course).

Therefore they had predicted we would likely struggle.

In those cases (as the Blackburn comment above) it's a useful marker to show if you are over or underperforming in the goal tally.

[Rileybobs]

Sure I saw a comment t last year where so eone had poi Ted out our xg:goals ratio was quite high Iin the Champ (eg. How many of those Benny wi ers were there which would be low xg) and therefor we scored a lot fro. Outside the box which are .uch harder to convert in the top league because you get less space (unless playing against is of course).

Therefore they had predicted we would likely struggle.

In those cases (as the Blackburn comment above) it's a useful marker to show if you are over or underperforming in the goal tally.

It’s a fair point about the number of ‘worldies’ we scored. But you would expect a team like Burnley to out-perform xG in the Championship because our players are better than the average opposition at that level, therefore will convert equivalent chances more than the average player at that level. The opposite goes for the Premier League.

I’m surprised so many people on here struggle with decimal points though.

[clarets1978]

Please can someone explain for me just what are Expected Goals and what is the point of the stat?

I listened to podcast years ago that hated Burnley under Sean Dyche because we always made a mockery of the apparently reliable xg stat. We'd come out some games having lost 4-0 according to xg yet in reality we'd have won 1-0

After hearing this I havent paid much attention to xg since!

[distortiondave]

Sure I saw a comment t last year where so eone had poi Ted out our xg:goals ratio was quite high Iin the Champ (eg. How many of those Benny wi ers were there which would be low xg) and therefor we scored a lot fro. Outside the box which are .uch harder to convert in the top league because you get less space (unless playing against is of course).

Therefore they had predicted we would likely struggle.

In those cases (as the Blackburn comment above) it's a useful marker to show if you are over or underperforming in the goal tally.

You wrote this like you were talking on a mobile phone whilst going through a tunnel.

[Mattster]

I listened to podcast years ago that hated Burnley under Sean Dyche because we always made a mockery of the apparently reliable xg stat. We'd come out some games having lost 4-0 according to xg yet in reality we'd have won 1-0

After hearing this I havent paid much attention to xg since!

Dyche doesn't make a mockery of xG unless you don't understand it (ie. Take it in complete isolation). Data like xG was/is a big factor in how Dyche sets up his team, especially his defence. He sets his defence up in a way where he isn't afraid to concede a lot of shots on goal, his teams encourage long range (low xG) chances and herd attackers into areas in the box where their only option is to shoot but the defenders are funneling the shot straight at the keeper. Those shots may be high xG due to where they're taken from (close to goal) in less sophisticated xG models and the sheer quantity Dyche is willing to allow, as the risk is mitigated, often put his sides on the wrong side of the xG statistic. More sophisticated models factor in defender and goalkeeper placement, whether the player has the ball fully under control etc. RE: Luton game, the reason the xG score was very close despite what the eye saw was because of these less sophisticated models used by free services at EFL level. If you look purely at where the shots were taken from without looking at positions of defenders, goalkeeper etc. our goals weren't from amazingly high chance positions - it's only when you note they were 1v1 chances, clear of the last man that the chance becomes bigger. Conversely Luton's two big chances (goal and crossbar) were very close to goal but only the goal was as good/better chance as our 3 open play goals as the header wasn't able to be directed due to Trafford (despite his mistake).

[distortiondave]

How do own goals work with the xG mechanic? e.g a ball across the box going in off the defender or Muric vs Brighton where there is no initial shot.

[quoonbeatz]

It’s for people who like stats rather than football.

[Mattster]

It’s for people who like stats rather than football.

quoon likes stats rather than football...

viewtopic.php?f=2&t=64809&p=1926827&hil ... t#p1926827

Or does that only apply to stats that have a decimal point in them?

[RVclaret]

How do own goals work with the xG mechanic? e.g a ball across the box going in off the defender or Muric vs Brighton where there is no initial shot.

Doesn’t get awarded anything from an xG perspective.

[quoonbeatz]

quoon likes stats rather than football...

viewtopic.php?f=2&t=64809&p=1926827&hil ... t#p1926827

Or does that only apply to stats that have a decimal point in them?

This might work if I’d been talking about xg, but good try my man, good try.

[RVclaret]

This might work if I’d been talking about xg, but good try my man, good try.

To justify a point you actually used a fairly meaningless stat. What if those shots on target were all from 25 yards and their 0 from 6 shots were from inside the box? You’d be better using xG next time.

[dsr]

One of the flaws in xG is that it doesn't take into account the circumstances of the whole move. For example, a few years back we scored direct from a corner at Southampton (when Ings inexplicably jumped out of the way). Initial thoughts were that Westwood took the corner, the keeper clawed it back off the line, Mee banged in the rebound from a yard. He would probably have had xG = 0.8 or 0.9. But what actually happened was that the keeper clawed the ball back from behind the line so xG was 0.01. If the keeper had made that save, it would have "earned" us almost a whole expected goal.

[Clive 1960]

maybe 0.5 of a goal is when ball is half way over goal line

[quoonbeatz]

To justify a point you actually used a fairly meaningless stat. What if those shots on target were all from 25 yards and their 0 from 6 shots were from inside the box? You’d be better using xG next time.

No, I used an absolute fact. If you think 0 out of 6 shots being on target means the keeper was tested then you have a lot to understand about football.

[Mattster]

This might work if I’d been talking about xg, but good try my man, good try.

Stats are stats. xG is just a stat, like shots and shots on target. It doesn't decide who wins a game, it simply provides further context to it. Saying one arbitrary stat is for people who like stats over football whilst the other aren't is just ridiculous.

[quoonbeatz]

Stats are stats. xG is just a stat, like shots and shots on target. It doesn't decide who wins a game, it simply provides further context to it. Saying one arbitrary stat is for people who like stats over football whilst the other aren't is just ridiculous.

I didn’t say that though. Not sure why you’re getting so wound up over what was clearly a tongue in cheek bit of prodding.

[Mattster]

I didn’t say that though. Not sure why you’re getting so wound up over what was clearly a tongue in cheek bit of prodding.

You said exactly that, though. Not sure why you're denying what you've said when everyone can see it

[quoonbeatz]

You said exactly that, though. Not sure why you're denying what you've said when everyone can see it

Not sure what you’re reading but everyone can definitely see that I didn’t say ‘one arbitrary stat is for people who like stats over football whilst others aren’t’.

I didn’t mention anything about other stats.

[Mattster]

Not sure what you’re reading but everyone can definitely see that I didn’t say ‘one arbitrary stat is for people who like stats over football whilst others aren’t’.

I didn’t mention anything about other stats.

And then when I brought up you mentioning other stats (shots and shots on target) you said it wasn't about those...

So it was about some stats but not others, as arbitrarily chosen by you. Yes?

[Winstonswhite]

Beyond the old adage that “there’s only one stat that counts”, xG is the most informative of stats as it takes in a multitude of factors rather than just shot on target or corners won, which I’ve always thought mean diddly squat.

[daveisaclaret]

Who's pretending!

You are pretending I think. Loads of good explanations and helpful links to your question and you are choosing to ask silly questions instead of learning the answer.