"Expected Goals" (xG)

[tomonred]

20 minutes ago, richinspain said:

 I think that our eyes are much better judges

Your eyes/my eyes/anyone's eyes are not better judges of a quality of chance than 100,000's of previous real life factual examples.

We're all too emotionally invested in it to make an accurate conclusion, because we are passionate fans.

In reality a chance where someone goes through on goal at a slight angle with a bit of pressure for a defender - we'd be screaming "he's got to score" etc - but in reality the quality of chance that it was, it might have historically only been scored 15-20% of the time (and as such is allocated a 0.15-0.2 xG rating)

[Will Hughes Hair]

13 minutes ago, tomonred said:

Your eyes/my eyes/anyone's eyes are not better judges of a quality of chance than 100,000's of previous real life factual examples.

We're all too emotionally invested in it to make an accurate conclusion, because we are passionate fans.

In reality a chance where someone goes through on goal at a slight angle with a bit of pressure for a defender - we'd be screaming "he's got to score" etc - but in reality the quality of chance that it was, it might have historically only been scored 15-20% of the time (and as such is allocated a 0.15-0.2 xG rating)

So hypothetically, the same chance falls to (let me choose an all time Rams favourite) Leon Best and Jamie Vardy.  The XG would be the same in both scenarios?

If so I'd suggest there is a fundamental flaw in using XG in a whole raft of scenarios.

[tomonred]

They are biased based on the standard you are playing at - e.g prem vs league two would have a slight different bias due to the quality of the historical data.  There's got to be a consistent metric for it to be gauged against to be able to draw any value.

Top level elite players will perform above the xG threshold consistently (Aguero/Kane etc because they are better finishers than the average). Looking at Vardy xG stats for example this season they are saying he should have 7 goals, but he has 12. He's performing above where a typical Premier League Striker would be expected to based on the chances he's had, and the stats are backing up the fact he is 'in form'

[richinspain]

30 minutes ago, tomonred said:

They are biased based on the standard you are playing at - e.g prem vs league two would have a slight different bias due to the quality of the historical data.  There's got to be a consistent metric for it to be gauged against to be able to draw any value.

Top level elite players will perform above the xG threshold consistently (Aguero/Kane etc because they are better finishers than the average). Looking at Vardy xG stats for example this season they are saying he should have 7 goals, but he has 12. He's performing above where a typical Premier League Striker would be expected to based on the chances he's had, and the stats are backing up the fact he is 'in form'

But the keepers are of the same standard as the striker, so a lower level keeper is expected to save more according to the weighting system? ?

[tomonred]

Some systems will be based on historical shot data for that league to make it appropriate to that league, so it will be factored into those final metrics yes

Regardless of the xG system, it will have the same core principle regardless of the league. The the values will just differ very slightly from league to league and very slightly from system to system

[DRBee]

26 minutes ago, tomonred said:

Some systems will be based on historical shot data for that league to make it appropriate to that league, so it will be factored into those final metrics yes

Regardless of the xG system, it will have the same core principle regardless of the league. The the values will just differ very slightly from league to league and very slightly from system to system

You are trying very hard but refining the data will not overcome the principal weaknesses of this bogus statistic. You cannot polish a lump of faeces.

Regardless of the limits you claim, commentators and idiots use 'expected goals' data to go far beyond what it can offer. Used this way  e.g. league tables based on predicted goals versus real results, you are expecting people who have watched a match ( emotional involvement or not) to cast aside all the other things we use to judge whether a match or performance is good or not in favour of a line on a graph which often 'flatlines' during times when we have seen excellent potentially match winning football.

[Sith Happens]

[Jram]

1 hour ago, DRBee said:

You are trying very hard but refining the data will not overcome the principal weaknesses of this bogus statistic. You cannot polish a lump of faeces.

Regardless of the limits you claim, commentators and idiots use 'expected goals' data to go far beyond what it can offer. Used this way  e.g. league tables based on predicted goals versus real results, you are expecting people who have watched a match ( emotional involvement or not) to cast aside all the other things we use to judge whether a match or performance is good or not in favour of a line on a graph which often 'flatlines' during times when we have seen excellent potentially match winning football.

I would completely believe a statistic over a fan any day because bias and emotional investment hugely skew your perception of what is happening 

Am I right in thinking that if you have a chance where almost 100% of players would score it that would be almost 1 on the xg? So could Prestons 1.7 be explained by them getting a score of 1 for the one chance where they missed basically an open goal? The shot before the offside which I think someone has screenshotted above. So that one chance (that they obv missed) bumped up their score 

[Dean (hick) Saunders]

God this thread is boring. Zzzz

[Comrade 86]

4 hours ago, tomonred said:

They are biased based on the standard you are playing at - e.g prem vs league two would have a slight different bias due to the quality of the historical data.  There's got to be a consistent metric for it to be gauged against to be able to draw any value.

Top level elite players will perform above the xG threshold consistently (Aguero/Kane etc because they are better finishers than the average). Looking at Vardy xG stats for example this season they are saying he should have 7 goals, but he has 12. He's performing above where a typical Premier League Striker would be expected to based on the chances he's had, and the stats are backing up the fact he is 'in form'

So in effect, what you've actually shown is how pointless the system really is. A wealth of data, squashed, washed and regurgitated to offer an estimate for Vardy's goal output that's not even in the same ballpark as the amount of goals he's actually scored. Other than to say he will score 'some goals', which we all know, you seem to be stating that the xG stats are useful because they show he is performing at a level above standard, but I could tell you that without watching a single game, let alone any statistical analysis.

Anyway, if you can't beat them... I've just crunched some big data myself and I reckon Vardy will score 23 goals this season. That'll be £5,000 please. My invoice is in the post. ?

[Comrade 86]

On 23/11/2019 at 20:54, LB_DCFC said:

Yeah I know. The recruitment team at Brentford, which is all built on statistical profiling of players are really doing a pointless and awful job. Not as if the best and most consistent recruitment team in the division have the most involvement of stats guys out of any club in the league or anything.

The metric of success being what? Player sell-on values?

I guess that's where we've got it all wrong, as I suspect play-off appearances or automatic promotion would be the ideal scenario for most clubs, Derby being a case in point. Perhaps though, an over reliance on statistical analysis is the reason why Brentford, despite their much hallowed recruitment, have failed to threaten either in recent years, despite a healthy P & L account. Does a stat that illustrates potential growth in value necessarily give clues as to how said recruits will piece together in a team? I'd venture that Brentford's recent campaigns would suggest not.

Moneyball isn't anything new really and serves a greater purpose where the selected 'tools' effectively play in isolation - baseball for instance. The variables across a 50+ game season of Championship and Cup football however, are rather harder to squash.

[LB_DCFC]

1 hour ago, 86 Schmokes & a Pancake said:

The metric of success being what? Player sell-on values?

I guess that's where we've got it all wrong, as I suspect play-off appearances or automatic promotion would be the ideal scenario for most clubs, Derby being a case in point. Perhaps though, an over reliance on statistical analysis is the reason why Brentford, despite their much hallowed recruitment, have failed to threaten either in recent years, despite a healthy P & L account. Does a stat that illustrates potential growth in value necessarily give clues as to how said recruits will piece together in a team? I'd venture that Brentford's recent campaigns would suggest not.

Moneyball isn't anything new really and serves a greater purpose where the selected 'tools' effectively play in isolation - baseball for instance. The variables across a 50+ game season of Championship and Cup football however, are rather harder to squash.

Success being the sheer quality of player they continue to find, pretty much all of which fit perfectly into their way of playing, plus the sell-on value. The only reason they are still in this division is the size of their club and the restriction in terms of revenue that entails. If they had the level of sponsorship we do and the gate receipts we do (their revenue is £17 million less than ours, for reference), they wouldn't need to sell their best player every year to balance the books as they currently do. They would've been able to accumulate a lot of talent and would've been promoted by now, I'm pretty sure of it. The use of stats doesn't then mean you then forgo all the other stuff as well, check on personality, style of player, and of course watching them. 

A squad with John Egan, Jota, Mepham, Maupay, Sawyers, Ryan Woods, Konsa plus their current team with Benrhama, Watkins, Jansson, Dasilva etc would be a top 2 squad. For a club like brentford, that is remarkable and all credit to their system. All those players were brought in for around a half of what we spent in the Clement season, and they are all better than any player we bought in those 2 transfer windows. 

The one thing you could argue they lacked in the last couple years is experience, but the quality of player they buy, and for the prices they buy them at, this makes their recruitment team by far and away the best in the division IMO, and if they continue this trajectory, they will get promoted at some point, whilst also making insane amounts of profit on player sales.

[rammieib]

On 23/11/2019 at 23:33, JfR said:

That's actually Ben Davies's shot immediately before the disallowed goal and he probably will be disappointed to have missed from here:

So he wasn't actually offside then?

Out of interest with XG, if a cross comes in and a player from one yard out miskicks at an open goal, would that count asan opportunity within XG?

I'm also surprised because the Waghorn goal must have had an XG of close to one being basically on the goal line?

[LB_DCFC]

14 hours ago, Tamworthram said:

Fair point. Still not a real job though. Makes you wonder though, if that's the route of their recruitment success, why others don't or can't adopt it more effectively. 

Others have, but not to the same extent and not as effectively. Their owner was a gambler who used the numbers to make his millions. The entire club is built to follow their system. In an environment such as football, where there are still a lot of people with ingrained beliefs contrary to reliance on stats , changing the entire culture of a club leaning in that direction is extremely difficult to do

[Comrade 86]

16 minutes ago, LB_DCFC said:

Success being the sheer quality of player they continue to find, pretty much all of which fit perfectly into their way of playing, plus the sell-on value. The only reason they are still in this division is the size of their club and the restriction in terms of revenue that entails. If they had the level of sponsorship we do and the gate receipts we do, they wouldn't need to sell their best player every year to balance the books as they currently do. They would've been able to accumulate a lot of talent and would've been promoted by now, I'm pretty sure of it. The use of stats doesn't then mean you then forgo all the other stuff as well, check on personality, style of player, and of course watching them. 

A squad with John Egan, Jota, Mepham, Maupay, Sawyers, Ryan Woods, Konsa plus their current team with Benrhama, Watkins, Jansson, Dasilva etc would be a top 2 squad. For a club like brentford, that is remarkable and all credit to their system. All those players were brought in for around a half of what we spent in the Clement season, and they are all better than any player we bought in those 2 transfer windows. 

The one thing you could argue they lacked in the last couple years is experience, but the quality of player they buy, and for the prices they buy them at, this makes their recruitment team by far and away the best in the division IMO, and if they continue this trajectory, they will get promoted at some point, whilst also making insane amounts of profit on player sales.

  

13 minutes ago, LB_DCFC said:

Success being the sheer quality of player they continue to find, pretty much all of which fit perfectly into their way of playing, plus the sell-on value. The only reason they are still in this division is the size of their club and the restriction in terms of revenue that entails. If they had the level of sponsorship we do and the gate receipts we do, they wouldn't need to sell their best player every year to balance the books as they currently do. They would've been able to accumulate a lot of talent and would've been promoted by now, I'm pretty sure of it. The use of stats doesn't then mean you then forgo all the other stuff as well, check on personality, style of player, and of course watching them. 

A squad with John Egan, Jota, Mepham, Maupay, Sawyers, Ryan Woods, Konsa plus their current team with Benrhama, Watkins, Jansson, Dasilva etc would be a top 2 squad. For a club like brentford, that is remarkable and all credit to their system. All those players were brought in for around a half of what we spent in the Clement season, and they are all better than any player we bought in those 2 transfer windows. 

The one thing you could argue they lacked in the last couple years is experience, but the quality of player they buy, and for the prices they buy them at, this makes their recruitment team by far and away the best in the division IMO, and if they continue this trajectory, they will get promoted at some point, whilst also making insane amounts of profit on player sales.

Yeah, hear this argument every year and it's actually getting quite dull. From a resale perspective, all good, but club success is not about P & L accounts, it's about delivering results on the pitch. Brentford continue to fall short and that's all there is to it. The revenue may hinder them I concede but Wigan amongst others have achieved it with modest revenues because they won enough games across a season. No analysis of that fact required! 

While I'd happily swap Brentford's scouting network for ours, I could say the same for half a dozen other sides in the div too. Our lot haven't exactly set th bar terribly high after all. Let's see how much the likes of Bogle, Lowe, Sibley and co are worth in a year or two though and then we can talk about what methodology benefits a team most.

[WoollyJumper]

Statistics now plays such a vital role for commentators and pundits with stuff to talk about and that's usually to pad to make enough content. It's easier and gives them a sense of knowledge. I think too much importance is placed on stats and they seem to get in the way of the facts.

[LB_DCFC]

1 minute ago, 86 Schmokes & a Pancake said:

Yeah, hear this argument every year and it's actually getting quite dull. From a resale perspective, all good, but club success is not about P & L accounts, it's about delivering results on the pitch. Brentford continue to fall short and that's all there is to it.

While I'd happily swap Brentford's scouting network for ours, I could say the same for half a dozen other sides in the div too. Our lot haven't exactly set th bar terribly high after all. Let's see how much the likes of Bogle, Lowe, Sibley and co are worth in a year or two though and then we can talk about what methodology benefits a team most.

Context matters though, dosen't it? Consistently getting top 10 finishes for a club like brentford is, in it's own way, an absolute roaring success. The club with the 2nd lowest revenue in the league (going off the 2017/18 accounts for reference) getting top 10 every year for about 5 years is no fluke in any way, and they show no sign of regressing at any point. They have become a mainstay within the conversation surrounding the upper echelons of the championship. 

By the way this doesn't mean i want us to shut down the academy or anything like they did. It was the right decision for them, considering all the competition from bigger clubs surrounding them. It wouldn't be the right decision for us to do the same imo.

[SaintRam]

Before reading this I really thought I was a stat guy.

[Comrade 86]

2 minutes ago, LB_DCFC said:

Context matters though, dosen't it? Consistently getting top 10 finishes for a club like brentford is, in it's own way, an absolute roaring success. The club with the 2nd lowest revenue in the league (going off the 2017/18 accounts for reference) getting top 10 every year for about 5 years is no fluke in any way, and they show no sign of regressing at any point. They have become a mainstay within the conversation surrounding the upper echelons of the championship. 

By the way this doesn't mean i want us to shut down the academy or anything like they did. It was the right decision for them, considering all the competition from bigger clubs surrounding them. It wouldn't be the right decision for us to do the same imo.

A fair point well made mate. As I said, I'd have their lot in tomorrow if it were feasible but I'd rather we do what we seem to be and build from the bottom with all age groups playing the same systems so that as they progress they fit in seamlessly, or as close as! Our academy has had kids join from Liverpool, Citeh and West Ham over the least year so necessarily the size of the club that governs the quality of intake, rather the facilities and the coaching staff.

Anyways, we digress. xG is a poo stat by any metric irrespective of the Bees rather more intelligent use of performance data. ?

[LB_DCFC]

2 minutes ago, 86 Schmokes & a Pancake said:

A fair point well made mate. As I said, I'd have their lot in tomorrow if it were feasible but I'd rather we do what we seem to be and build from the bottom with all age groups playing the same systems so that as they progress they fit in seamlessly, or as close as! Our academy has had kids join from Liverpool, Citeh and West Ham over the least year so necessarily the size of the club that governs the quality of intake, rather the facilities and the coaching staff.

Anyways, we digress. xG is a poo stat by any metric irrespective of the Bees rather more intelligent use of performance data. ?

Bigger club was probably the wrong phrase to use. Clubs with more established reputation in terms of facilities and coaching would be better.

Good to see not every internet discussion has to end with both people hurling abuse at each other. Quite nice actually, I don't quite know what to do ?