Latest xG table

I'd be keen to know this too. Have asked before and not got an answer.

My issue with it is that the table in the OP is (I assume) the prediction as we stand. As soon as any result changes what the prediction was at that time, the prediction changes again.

It's not relevant at that point.

It's just like a bookies odds being reflected in changes in sporting events.

Can see why it's fun for some people to discuss but if it's only "right until a random / unexpected act changes it" it's nothing more than that.
That’s true for any forecast or prediction, which are used in any huge industry.

House market looks strong until the cost of living crisis. Stocks looked promising until covid. My works fully covered until people take sick days.

The problem with xG is more with people not understanding it or applying it correctly. Of course it isn’t going to be correct all the time, it would have predicted Rangers to beat Motherwell at Ibrox but it has its purpose and that’s why just about every sports team on the planet are now using it or something similar.
 
That’s true for any forecast or prediction, which are used in any huge industry.

House market looks strong until the cost of living crisis. Stocks looked promising until covid. My works fully covered until people take sick days.

The problem with xG is more with people not understanding it or applying it correctly. Of course it isn’t going to be correct all the time, it would have predicted Rangers to beat Motherwell at Ibrox but it has its purpose and that’s why just about every sports team on the planet are now using it or something similar.

Yup, I know why bookies and the likes use it. It helps narrow down a field and provides information that helps many sporting points.

People posting it on here though tend to do so as if it's a nailed on fact though.

It might indicate things right now but it'll likely be very different come 3pm on Sunday.

I think that's a lot of people's issue with the discussion in general. It'll be bang on...until it's not.

From a discussion point of view here with no real basis other than the fact presented itself, it's no more use than looking at the bookie's guess just now on who'll win the league and assume that's deffo right.
 
I'd be keen to know this too. Have asked before and not got an answer.

My issue with it is that the table in the OP is (I assume) the prediction as we stand. As soon as any result changes what the prediction was at that time, the prediction changes again.

It's not relevant at that point.

It's just like a bookies odds being reflected in changes in sporting events.

Can see why it's fun for some people to discuss but if it's only "right until a random / unexpected act changes it" it's nothing more than that.
Of course it's only a prediction based on available data - there's nothing else to base it on.

There are lots of papers showing that it predicts future results betteer than "traditional" stats. (Example: https://www.americansocceranalysis....t-is-xg-the-best-predictor-of-future-results_.

Think the problem is that some people can overplay how well they can predict future results. It's an attempt to model something which is inherently difficult to predict - it's never going to be perfect.
 
I'd be keen to know this too. Have asked before and not got an answer.

My issue with it is that the table in the OP is (I assume) the prediction as we stand. As soon as any result changes what the prediction was at that time, the prediction changes again.

It's not relevant at that point.

It's just like a bookies odds being reflected in changes in sporting events.

Can see why it's fun for some people to discuss but if it's only "right until a random / unexpected act changes it" it's nothing more than that.

Why do you have an issue if a prediction changes the more information it gets? That's the point, it takes all the information it has at the time and makes a prediction based on that. It would be nonsense if it just ignored known results and just stubbornly stuck to what it predicted at the beginning of the season - that's human behaviours (weaknesses).

The model mentioned in the OP has been very accurate over the last few seasons and predicted postecoglou's first season win while most were still laughing at how bad they were.
 
Of course it's only a prediction based on available data - there's nothing else to base it on.

There are lots of papers showing that it predicts future results betteer than "traditional" stats. (Example: https://www.americansocceranalysis....t-is-xg-the-best-predictor-of-future-results_.

Think the problem is that some people can overplay how well they can predict future results. It's an attempt to model something which is inherently difficult to predict - it's never going to be perfect.


Why do you have an issue if a prediction changes the more information it gets? That's the point, it takes all the information it has at the time and makes a prediction based on that. It would be nonsense if it just ignored known results and just stubbornly stuck to what it predicted at the beginning of the season - that's human behaviours (weaknesses).

The model mentioned in the OP has been very accurate over the last few seasons and predicted postecoglou's first season win while most were still laughing at how bad they were.

The bolded bit is the exact point I'm making in post #52. People post the stats as if they're facts that can't / won't change.

It's nothing more than a tool for interesting discussion without knowing all the point that have got it to reach that conclusion as things stand.

If Butland gets injured tomorrow in training and is out for the season, the prediction will adjust and the entire thing is up in the air.

As things stand for this exact moment, the prediction might be correct (or not) and it will change minorly over very slight things and likely majorly come next week.

People fixate on it like it's bang on and can't change. It's not that I've really got issue with the information.

I've got an issue with the people who ignore that very fact and present it as if it's set in stone.
 
The bolded bit is the exact point I'm making in post #52. People post the stats as if they're facts that can't / won't change.

It's nothing more than a tool for interesting discussion without knowing all the point that have got it to reach that conclusion as things stand.

If Butland gets injured tomorrow in training and is out for the season, the prediction will adjust and the entire thing is up in the air.

As things stand for this exact moment, the prediction might be correct (or not) and it will change minorly over very slight things and likely majorly come next week.

People fixate on it like it's bang on and can't change. It's not that I've really got issue with the information.

I've got an issue with the people who ignore that very fact and present it as if it's set in stone.

The very first word of the thread title shows that it's changeable and not 'set in stone'.
 
Can you show any examples of people acting as if that xG table is set in stone.

I’ve yet to see anyone who is in favour of xG claiming that.

Go and look up any thread on here previously posted and you'll see people commenting on it in such a fashion.
 
Yup, I know why bookies and the likes use it. It helps narrow down a field and provides information that helps many sporting points.

People posting it on here though tend to do so as if it's a nailed on fact though.

It might indicate things right now but it'll likely be very different come 3pm on Sunday.

I think that's a lot of people's issue with the discussion in general. It'll be bang on...until it's not.

From a discussion point of view here with no real basis other than the fact presented itself, it's no more use than looking at the bookie's guess just now on who'll win the league and assume that's deffo right.
it works very well for a number of different things. Scouts use it for player information, being able to compare. There was a point last season Haaland and Darwin had the same xG roughly and Haaland had about 15 more goals showing his ruthlessness and Darwin’s wastefulness. Same can be done for any position on the pitch.

Same for team level analysis, the prediction correctly suggested that Ange’s Celtic were playing better than their points suggested and we were playing worse, we ended up falling behind. Similar predictions this year.

It can never be perfect and that’s the case for any model and it will change regardless of what happens this weekend as it should because one or both of the teams will have lost points and it will recalculate.
 
Yeah the notion that past performance can dictate a trend in future performance is ridiculous. Everything is random.
Unless it takes injuries, weather conditions, refereeing mistakes and every other variable into consideration, then it really is random. I do take an interest in it, but I often look at the xG after I've watched a game and sometimes it's just bizarre. I can't take it too seriously. I know some take it as gospel, so to each his own.
 
It's really not. The very point comes up on every thread.

In what way? Are people celebrating as if the league is won because we're currently top of the xG?

If you can't point to any concrete examples can you at least say how people treat it as if it's set in stone?

It's a tool to predict future point accumulation based on the quality of chances teams are creating/conceding and it has proved relatively accurate. It changes throughout the season and people who like it all recognise that fact.
 
I think the XG from October when the Scum were clear at the top showed it was a bit of a false position.

Maybe there is something in this data stuff after all.
 
In what way? Are people celebrating as if the league is won because we're currently top of the xG?

If you can't point to any concrete examples can you at least say how people treat it as if it's set in stone?

It's a tool to predict future point accumulation based on the quality of chances teams are creating/conceding and it has proved relatively accurate. It changes throughout the season and people who like it all recognise that fact.

Not all they don't.

However, at this point, you're simply spoiling for an argument so I'll leave you there...
 
The last post I seen on xG I asked somebody to explain it to me.

It made sense... now I'm seeing xG being negative and I'm confused again.

Need another explainer
The stat in the table is xGD not xG. GD meaning goal difference. So it’s their xG minus the xG teams get when playing against them.
 

I think what I'd find interesting would be the prediction at the start of the season compared to the actual end result and how many adjustments were made during all of the modelling of the season to react to change.

The likes of the prediction on Gio...it comes in September the day after the 4-0 scudding at The Piggery What was it saying 2 months before and what was the catalyst for it's huge change. Was it simply after that result that the big gap appeared?
 
Why do you have an issue if a prediction changes the more information it gets? That's the point, it takes all the information it has at the time and makes a prediction based on that. It would be nonsense if it just ignored known results and just stubbornly stuck to what it predicted at the beginning of the season - that's human behaviours (weaknesses).

The model mentioned in the OP has been very accurate over the last few seasons and predicted postecoglou's first season win while most were still laughing at how bad they were.
I’ve no problem with this but, the xg stat seems to be very “moveable”.
For example, the chance that Dessers missed on Saturday, also Tavernier’s penalty miss, do they increase the chances of winning the league?
Lawrence’s decision to pass rather than shoot, how does that influence the stat?
I’m asking because, with these three misses, Rangers still won.
 
Last edited:
I think what I'd find interesting would be the prediction at the start of the season compared to the actual end result and how many adjustments were made during all of the modelling of the season to react to change.

The likes of the prediction on Gio...it comes in September the day after the 4-0 scudding at The Piggery What was it saying 2 months before and what was the catalyst for it's huge change. Was it simply after that result that the big gap appeared?

Like this sort of thing where he shows the full season progression?

I see what you mean about predicting from before the start of the season - I think his model needs some games to have been played first potentially.
 

Like this sort of thing where he shows the full season progression?

I see what you mean about predicting from before the start of the season - I think his model needs some games to have been played first potentially.

Yes, exactly like that. Cheers. It helps to see the accuracy when presented like that.

If they added in when the prediction was made compared to the actual match day, it'd help too.
 
I’ve no problem with this but, the xg stat seems to be very “moveable”.
For example, the chance that Dessers missed on Saturday, also Tavernier’s penalty miss, do they increase the chances of winning the league?
Lawrence’s decision to pass rather than shoot, how does that influence the stat?
I’m asking because, with these three misses, Rangers still won.

Having a higher xGD seems to increase the chances of winning the league, yes. Presumable because you point out, chances will be missed so the more chances of higher quality you create the more chance you have of still getting the 3 points.

If our xG was lower then those misses you highlight are more likely to cost us points.
 
Problem with the model is that it penalises us for having a game in hand - so the predicted gap is smaller than if Celtic hadn’t yet played Livi.
 
Newcastle XG tonight 2.67 - scored 1
Forest XG tonight 1.30 - scored 3
Portsmouth XG tonight - 0.85 - scored 2

It doesn’t even out
You realise that picking other isolated games that suit your narrative doesn’t increase the sample size for what you’re trying to prove here?

I notice you chose two EPL games and ignored the other three. Any reason for this?
 
Newcastle XG tonight 2.67 - scored 1
Forest XG tonight 1.30 - scored 3
Portsmouth XG tonight - 0.85 - scored 2

It doesn’t even out
You’ve quoted 3 individual games again, which doesn’t address the point of it correcting and evening out over time. It will never be perfect because Football is fundamentally unpredictable, but it’s as good as you can get and anomalies almost always correct over time. For example, Hibs have an xGD of -5 and an actual of -6. That is ridiculously accurate after 31 games.
 
Last edited:
The bolded bit is the exact point I'm making in post #52. People post the stats as if they're facts that can't / won't change.

It's nothing more than a tool for interesting discussion without knowing all the point that have got it to reach that conclusion as things stand.

If Butland gets injured tomorrow in training and is out for the season, the prediction will adjust and the entire thing is up in the air.

As things stand for this exact moment, the prediction might be correct (or not) and it will change minorly over very slight things and likely majorly come next week.

People fixate on it like it's bang on and can't change. It's not that I've really got issue with the information.

I've got an issue with the people who ignore that very fact and present it as if it's set in stone.

If Butland gets injured tomorrow it won't change the xG. It's related to the quality of chance created by a team.
 
I think what I'd find interesting would be the prediction at the start of the season compared to the actual end result and how many adjustments were made during all of the modelling of the season to react to change.

The likes of the prediction on Gio...it comes in September the day after the 4-0 scudding at The Piggery What was it saying 2 months before and what was the catalyst for it's huge change. Was it simply after that result that the big gap appeared?

There can't be an xG prediction at the start of the season. There needs to be games played to have the data to extrapolate those performances across the season.

Really needs to be a good number of games played as well in order to get enough data, probably a full round of matches.
 
Newcastle XG tonight 2.67 - scored 1
Forest XG tonight 1.30 - scored 3
Portsmouth XG tonight - 0.85 - scored 2

It doesn’t even out

This is just continuing to show you don't understand it.

Newcastle batter Everton for most of the game and missed numerous good chances.

If a team has a higher xG than goals scored its showing that there is a problem converting chances.

If they have a lower xG than goals scored, then they are more than likely riding their luck and results will take a down turn.

Low xg and low goals means they are just shit.

It's more a tool for coaches to identify what they need to work on rather than anything else
 
xG was correct for 8/10 EPL teams tonight. And that’s without the benefit of evening out over time. Incredible to know that and plough ahead with a flawed argument
So he picked out the only two teams from ten that they weren’t accurate for to try and prove his point? :D
 
Yes, exactly like that. Cheers. It helps to see the accuracy when presented like that.

If they added in when the prediction was made compared to the actual match day, it'd help too.
It’s a dataset for each season based on the data available for that season.

So basically it gets more accurate the more games are played.

If you done this after only two or three games I don’t think it would be all that valuable
 
You’ve quoted 3 individual games again, which doesn’t address the point of the it correcting and evening out over time. It will never be perfect because Football is fundamentally unpredictable, but it’s as good as you can get and anomalies almost always correct over time. For example, Hibs have an xGD of -5 and an actual of -6. That is ridiculously accurate after 31 games.
[/QUOTE]
Aye but Sutton united won 3-0 today an xG of 1.45
 
But if Dessers doesn't take the chance creation then it means hee haw.
No different from AI predicting the league.
;)
Dessers I believe has a higher number of goals than his xG, more so since the turn of the year, as he scores the more difficult chances with really low xG.

But way to go trying to derail another thread slagging our players.
 
You’d think some users on this founded XG with how particular they are about it.

I hope Rangers win the XG league, it’ll make me ever so happy
 
You’d think some users on this founded XG with how particular they are about it.

I hope Rangers win the XG league, it’ll make me ever so happy
Subtext: “I don’t have a fucking clue what I’m talking about, so instead of admitting it I’ll try a weak dig at people who do”

By the way, just because we understand something doesn’t mean we place a lot of importance in it.
 
A lack of understanding of probability suggests bigger problems with cognitive thinking.

Superscoreboard proves this nightly.
 
Back
Top