Attention: Please take a moment to consider our terms and conditions before posting.

Expected Goals

Just read this excellent book on how important data analysis has become in football. Clubs like Brighton, Brentford and the top 6 in the Premiership have changed the way they think and use the data available now on every single action in a game to make decisions on everything from player recruitment to how they set up the team.
The ‘high press’ we’ve adopted under Jones is a classic example of a club using data to create goal opportunities and I understand that players now use it to choose their next club!
Does Nathan Jones include this data in his decisions?
Does Charlton have data analysts to provide this info?
I haven’t contacted the club directly to ask as the secrecy involved probably means they wouldn’t divulge their process but does anyone on here have knowledge of our clubs approach?
PS I vividly remember the abuse the mad Belgian received after disclosing his use of the student maths expert so I’m wearing my tin helmet preparing for the storm!
«1

Comments

  • edited August 12
    I get that XG says percentage of shots from a particular grid point are expected to result in goals but does that take into account if just getting a clear strike at ball sitting nicely on ground or if ball is fizzed into that grid spot at knee height or if player is tightly marked etc ? 
  • Apparently, yes!
    These data companies have teams of people grading every aspect of passing, movement, tackling, heading etc etc on every game.
    They provide incredibly detailed data which then needs data analysts.to interpret
  • Does it take into account the player in that position. A top notch striker or a crap defender. 
  • Does it take into account the player in that position. A top notch striker or a crap defender. 
    Or the goalkeeper they are up against? 
  • Yes it takes into account all of those things, and more. It also takes into account the time in the game, the weather, temperature and humidity, how long the player has been on the pitch for and even if there's any wind factor to take into account. 

    The data behind all these stats is phenomenal, the amount that they track through the sports vests they wear (training bras) is absolutely crazy, utilising a mixture of sensors to track movements via GPS, speed, how much power they using and things like heart rate. 

    I have a cousin that works in football data for West Ham and the stuff they cover in data analysis now is mind boggling 
  • The issue with xG is how it's been presented to the public. It isn't really a good thing to use on a game by game basis, you can tell that by eye. xG is good over a span of 10+ games to see if there are any trends but really it's a surface level stat before you dive in deeper
  • fenaddick said:
    The issue with xG is how it's been presented to the public. It isn't really a good thing to use on a game by game basis, you can tell that by eye. xG is good over a span of 10+ games to see if there are any trends but really it's a surface level stat before you dive in deeper
    This is the issue with all data, analysis and stats. They are incredibly technical and so much goes into it. They are designed to be used in certain ways but media present them differently and the public misunderstand their uses and suddenly conclusions are jumped to based on wrong assumptions and peoples views are formed on that basis!

    COVID Showed this up massively. The average person is numerate but not data literate. 
  • fenaddick said:
    The issue with xG is how it's been presented to the public. It isn't really a good thing to use on a game by game basis, you can tell that by eye. xG is good over a span of 10+ games to see if there are any trends but really it's a surface level stat before you dive in deeper
    This is the issue with all data, analysis and stats. They are incredibly technical and so much goes into it. They are designed to be used in certain ways but media present them differently and the public misunderstand their uses and suddenly conclusions are jumped to based on wrong assumptions and peoples views are formed on that basis!

    COVID Showed this up massively. The average person is numerate but not data literate. 
    Agreed.

    I actually think individual xG/xA/xSaves is better than for teams. Much easier to pinpoint strengths and weaknesses that way
  • Football is a very simple game to understand. 

    Score more than the opposition and you win.

    Score less than the opposition and you lose.

    Everything else is just noise. 
    I used to think that until I saw coaching and team talks up close. There's a lot more going on on the pitch then any fan will know. 

    I do think XG and the like is a bit wanky, and feels like the Americans and their fetish for stats. It's just one factor in a plethora and can be used to inform, but not make direct decisions. It can reinforce or challenge an opinion, but should never be used in isolation to form an opinion on a player. 
  • Sponsored links:


  • It's good when used by clubs to analyse chances created or allowed in games and see long-term trends in performance. 

    It's bad when used to cherry pick single-game results and claim the score isn't 'real'. 
  • The club does have data analysts and has done so for a while.

    However, how well they were used under the previous regimes and this one is debatable.  But they club has said that it is a department they have improved.

    Data is just raw material, it is how it is how it is understood and then used that matter and that applies to football and many other fields.

    For example, all the players are tested at the start of their training days, that data is used to frame their training for the day.  That should help avoid injuries.

    As said, XG is a bit of a blunt tool and seemed to be all the rage a few seasons ago among fans who thought it revealed the secrets of the universe when it didn't as they were only getting one headline figure on one game not all the underlying information or trends.  

    It now appears to have gone out of fashion, just as possession percentages have to a lesser extent.
  • sam3110 said:
    Yes it takes into account all of those things, and more. It also takes into account the time in the game, the weather, temperature and humidity, how long the player has been on the pitch for and even if there's any wind factor to take into account. 

    The data behind all these stats is phenomenal, the amount that they track through the sports vests they wear (training bras) is absolutely crazy, utilising a mixture of sensors to track movements via GPS, speed, how much power they using and things like heart rate. 

    I have a cousin that works in football data for West Ham and the stuff they cover in data analysis now is mind boggling 
    For an oldie like me, this is merely gobbledygook. 

    But there is one aspect mentioned in sam's post that caught my eye.

    " The wind factor..."

    Forget all the stats, data, humidity etc etc. It's really not that complicated. 

    Just concentrate on one single factor that those of a certain age have used successfully for decades to increase their speed & agility.

    Two cans of baked beans per player an hour or so before kick off ..then stand back to see the results....

    ....as long as the lads go nowhere near a naked flame of course.

    Thank you Mr Heinz.
  • Does it take into account the player in that position. A top notch striker or a crap defender. 
    Believe there are different versions meaning depending which company you use you may get slightly different results.

    I don’t think who the player is gets taken into account as the point is you’re able to compare players vs the average. A top striker might score several more than xG while a poor one will be below xG, that shows you who the more effective goalscorer.

    If you used who the player was then that brings opinion into it when the point of it is to compare players based on the data.
  • The book has an interesting section on what happened with our club when ‘The Spivs’ were in charge.
    One of the passionate early adopters of data analysis in football realised that most ‘Old School’ managers wouldn’t fully implement the data results so he wanted to buy a club and run it completely based on data.
    The perfect club he wanted was . . . Charlton Athletic . . . 
    But he had a problem, he couldn’t find out who actually owned the club. Neither the Spivs or the mysterious Mr Cash actually owned it and it wasn’t until a court case some years later that it was decided.
  • The club does have data analysts and has done so for a while.

    However, how well they were used under the previous regimes and this one is debatable.  But they club has said that it is a department they have improved.

    Data is just raw material, it is how it is how it is understood and then used that matter and that applies to football and many other fields.

    For example, all the players are tested at the start of their training days, that data is used to frame their training for the day.  That should help avoid injuries.

    As said, XG is a bit of a blunt tool and seemed to be all the rage a few seasons ago among fans who thought it revealed the secrets of the universe when it didn't as they were only getting one headline figure on one game not all the underlying information or trends.  

    It now appears to have gone out of fashion, just as possession percentages have to a lesser extent.
    Thanks for the update, and I do remember that day when your ‘moniker’ the statue, was incorporated into our heritage.
    So the club does use data analysis, it would be very interesting to find out how it is implemented on a regular basis.
  • The club does have data analysts and has done so for a while.

    However, how well they were used under the previous regimes and this one is debatable.  But they club has said that it is a department they have improved.

    Data is just raw material, it is how it is how it is understood and then used that matter and that applies to football and many other fields.

    For example, all the players are tested at the start of their training days, that data is used to frame their training for the day.  That should help avoid injuries.

    As said, XG is a bit of a blunt tool and seemed to be all the rage a few seasons ago among fans who thought it revealed the secrets of the universe when it didn't as they were only getting one headline figure on one game not all the underlying information or trends.  

    It now appears to have gone out of fashion, just as possession percentages have to a lesser extent.
    XG had its time as the new big thing.  Lots of noise was created by people who quoted and requoted it for fear of sounding uninformed or out of date.
    Statistical analysis can be useful of course.  Some sad sacks really like stats for the sake of them.  Mostly they are just data.  They can only be useful when they amount to information.  A team's XG in any number of games could be 10 or 15 - don't mean they won any games or even scored a single goal.
    Dunno what Wigan or Charlton's XG was on Saturday.  The only stat that matters is the final score.
    No manager or fan for that matter needs XG to tell him that a team might be creating loads of chances but scoring few goals.
    Having and XG close to zero game after game doesn't mean you've lost or failed to score - Saturday's game being the perfect example.m_2 said:
    The book has an interesting section on what happened with our club when ‘The Spivs’ were in charge.
    One of the passionate early adopters of data analysis in football realised that most ‘Old School’ managers wouldn’t fully implement the data results so he wanted to buy a club and run it completely based on data.
    The perfect club he wanted was . . . Charlton Athletic . . . 
    But he had a problem, he couldn’t find out who actually owned the club. Neither the Spivs or the mysterious Mr Cash actually owned it and it wasn’t until a court case some years later that it was decided.
    Didn't the bods that bought Brentford do so with the intention of relying heavily on some statistical modelling idea?  They did away with the academy.
    However long they used it for, can't say it wasn't successful.
  • The club does have data analysts and has done so for a while.

    However, how well they were used under the previous regimes and this one is debatable.  But they club has said that it is a department they have improved.

    Data is just raw material, it is how it is how it is understood and then used that matter and that applies to football and many other fields.

    For example, all the players are tested at the start of their training days, that data is used to frame their training for the day.  That should help avoid injuries.

    As said, XG is a bit of a blunt tool and seemed to be all the rage a few seasons ago among fans who thought it revealed the secrets of the universe when it didn't as they were only getting one headline figure on one game not all the underlying information or trends.  

    It now appears to have gone out of fashion, just as possession percentages have to a lesser extent.
    It might not be true and no doubt someone would know, but I thought in recent years the more possession Charlton had the worse our results.
    I'm pretty sure on Saturday Wigan had far more possession.
  • edited August 12
    If you think xG is interesting then you should do some reading on Expected Threat. 

    Expected Goals takes us from goals, back to shots as an indicator of performance. Expected Threat takes the concept one step further and starts to look at possession, passing and getting into dangerous areas with the ball to hurt the opposition and can assign values to that, even if no shot was taken.
  • Sponsored links:


  • What was the XG from Saturdays humdinger 
  • What was the XG from Saturdays humdinger 
    Wigan 1.05
    Charlton 0.42
  • What was the XG from Saturdays humdinger 
    Wigan 1.05
    Charlton 0.42
    Looking at it in more detail. The highest Wigan chance was the Stones header near the end 0.31. The biggest Charlton xg was the opportunity for Docherty 0.28. Our goal through Lloyd Jones was only 0.1 xg. The vast majority of Wigan's 'chances' were 0.1 or under. The Aasgaard chance for example only added 0.03 xg.
  • Billy_Mix said:
    The club does have data analysts and has done so for a while.

    However, how well they were used under the previous regimes and this one is debatable.  But they club has said that it is a department they have improved.

    Data is just raw material, it is how it is how it is understood and then used that matter and that applies to football and many other fields.

    For example, all the players are tested at the start of their training days, that data is used to frame their training for the day.  That should help avoid injuries.

    As said, XG is a bit of a blunt tool and seemed to be all the rage a few seasons ago among fans who thought it revealed the secrets of the universe when it didn't as they were only getting one headline figure on one game not all the underlying information or trends.  

    It now appears to have gone out of fashion, just as possession percentages have to a lesser extent.
    XG had its time as the new big thing.  Lots of noise was created by people who quoted and requoted it for fear of sounding uninformed or out of date.
    Statistical analysis can be useful of course.  Some sad sacks really like stats for the sake of them.  Mostly they are just data.  They can only be useful when they amount to information.  A team's XG in any number of games could be 10 or 15 - don't mean they won any games or even scored a single goal.
    Dunno what Wigan or Charlton's XG was on Saturday.  The only stat that matters is the final score.
    No manager or fan for that matter needs XG to tell him that a team might be creating loads of chances but scoring few goals.
    Having and XG close to zero game after game doesn't mean you've lost or failed to score - Saturday's game being the perfect example.m_2 said:
    The book has an interesting section on what happened with our club when ‘The Spivs’ were in charge.
    One of the passionate early adopters of data analysis in football realised that most ‘Old School’ managers wouldn’t fully implement the data results so he wanted to buy a club and run it completely based on data.
    The perfect club he wanted was . . . Charlton Athletic . . . 
    But he had a problem, he couldn’t find out who actually owned the club. Neither the Spivs or the mysterious Mr Cash actually owned it and it wasn’t until a court case some years later that it was decided.
    Didn't the bods that bought Brentford do so with the intention of relying heavily on some statistical modelling idea?  They did away with the academy.
    However long they used it for, can't say it wasn't successful.
    Exactly, and they still use data in everything they do. When those players get out on the grass they know the percentages of what will work and what won’t. No long shots, high pressing, routines at set pieces, little old Brentford in the Premiership. Lessons to be learned there . . . And then there’s Brighton, Luton etc etc
  • The club does have data analysts and has done so for a while.

    However, how well they were used under the previous regimes and this one is debatable.  But they club has said that it is a department they have improved.

    Data is just raw material, it is how it is how it is understood and then used that matter and that applies to football and many other fields.

    For example, all the players are tested at the start of their training days, that data is used to frame their training for the day.  That should help avoid injuries.

    As said, XG is a bit of a blunt tool and seemed to be all the rage a few seasons ago among fans who thought it revealed the secrets of the universe when it didn't as they were only getting one headline figure on one game not all the underlying information or trends.  

    It now appears to have gone out of fashion, just as possession percentages have to a lesser extent.
    It might not be true and no doubt someone would know, but I thought in recent years the more possession Charlton had the worse our results.
    I'm pretty sure on Saturday Wigan had far more possession.
    Very true, our players were going backwards and sideways whereas with the right coaching the ball goes forwards. The data will show which opposition players press and which ones don’t and the plan works around the space going forwards
  • Scoham said:
    Does it take into account the player in that position. A top notch striker or a crap defender. 
    Believe there are different versions meaning depending which company you use you may get slightly different results.

    I don’t think who the player is gets taken into account as the point is you’re able to compare players vs the average. A top striker might score several more than xG while a poor one will be below xG, that shows you who the more effective goalscorer.

    If you used who the player was then that brings opinion into it when the point of it is to compare players based on the data.
    Spot on and the result is that most premiership clubs now own a data modelling company as they couldn’t trust sharing a supplier with the competition so they bought them. Very interested in how our club goes about it and how much Nathan uses it.
  • There’s quite a bizarre desperation to dismiss xG amongst football fan.  It’s not meaningless, the cream of the crop employ all sorts of analysts these days, it’s as important as ever.  

    It’s not supposed to give you an idea of how a game should have gone.  It’s supposed to give you an idea of how the next one might go if you don’t tweak something.  

    The more info you have at your disposal the better.
    Agreed, and the calculated table based on expected points is interesting too.

    When a team is much higher up the table than the "xG calculated table", it does indicate a slightly false position, where a team has been getting wins in game they probably should have lost, which is unlikely to continue. Similarly, if a team is below its xG position, then things are likely to turn around soon, without having to change the manager or drastically changing the team or tactics. 
  • edited August 12
    What was the XG from Saturdays humdinger 
    Wigan 1.05
    Charlton 0.42
    Looking at it in more detail. The highest Wigan chance was the Stones header near the end 0.31. The biggest Charlton xg was the opportunity for Docherty 0.28. Our goal through Lloyd Jones was only 0.1 xg. The vast majority of Wigan's 'chances' were 0.1 or under. The Aasgaard chance for example only added 0.03 xg.
    I saw the same numbers in the FotMob app (which uses Opta data) earlier today but I’ve just gone back into the app and it appears that the xG has been recalculated… 0.78 for Wigan, 0.46 for Charlton.
  • What was the XG from Saturdays humdinger 
    Wigan 1.05
    Charlton 0.42
    Looking at it in more detail. The highest Wigan chance was the Stones header near the end 0.31. The biggest Charlton xg was the opportunity for Docherty 0.28. Our goal through Lloyd Jones was only 0.1 xg. The vast majority of Wigan's 'chances' were 0.1 or under. The Aasgaard chance for example only added 0.03 xg.
    I saw the same numbers in the FotMob app (which uses Opta data) earlier today but I’ve just gone back into the app and it appears that the xG has been recalculated… 0.78 for Wigan, 0.46 for Charlton.
    So the forecasting tool for future performance has just killed my first game win optimism. 
Sign In or Register to comment.

Roland Out Forever!