‘XG’

[blades95]

Do all companies and individuals share the same database to calculate XG? Or does everybody compile their own?

I believe they'd use the same database however different people use a different formula to calculate xG

 

[SwissBlade]

Like it or not, the use of stats like this, as well as things like heat maps is part of the future of football coaching at the top level. It doesn't replace the basics, but it can add value if used intelligently

I was in Houston about 6-7 years back and I had a couple of days working in the hotel room. In that time I had ESPN on and they were going through the draft/ draught for basketball, think it was college leagues or something, but also ice hockey.

In both they were having these debates using some mad stats which didn’t make much sense to me but was clearly aimed at proving points. It was really interesting and I couldn’t really see how it could be applied to football when we used so little

Now with gps vests, heat maps etc and other info produced from these data sets it gives a new insight.

Things like xg are theoretical to an extent but still useful.

I’m going way off topic now. But now that this data can be used in real time it does change things for managers and coaches.

As I said before, Clough was using the gps vests in training but I don’t think we were allowed to use them in games until Adkins came in and I’m not sure we had the equipment in the ground at the lane at that time to really benefit from it. So it’s become much more noticeable under wilder.

There is no hiding anymore.

I watched an interesting documentary on prime last week, about ultra runners doing a two peak 140mile run in the himalayas. How the athletes in the race were monitored during the race and advised to slow down before they fell down. It’s a long way from footballers doing 90 mins and their methods were different but sports science nowadays allows us to be able to know when a player is hitting the red before they do

 

[Fanny magnet]

No, it isn’t. It’s absolute rubbish.

That's just because you don't understand it

 

[grafikhaus]

xG is more scientific and meaningful than shots on target because, as you say, one from 40 yards that rolls into the arms is counted the same as a tap in on the line. xG takes these factors into account. I don’t mind it, I wouldn’t obsessively use them but it’s a decent indication of whether you’re doing the right things or not, ie creating good chances for your strikers

I'd say this is a decent indication of whether you're doing the right things or not? (Seriously, if your strikers are shit and the transfer window has closed, what can you do about it?)

Anyway, get back to yer book on Billy.

Like it or not, the use of stats like this, as well as things like heat maps is part of the future of football coaching at the top level. It doesn't replace the basics, but it can add value if used intelligently

Heat maps and GPS vests are very useful. Is a player tiring and needs subbing? Despite what he said, have we rushed this guy back from injury too soon? etc.

But this xg stuff smacks of 'garbage in, garbage out'.

 

[JD1889]

I think it's really interesting. Is there an up to date list based on it for the championship? Interested in our xg v the teams we have yet to play.

Found these from Blades Analytics tweets.

 

[BushBlade]

What I’d find interesting is to know how much credibility the bookies give these stats. If they’re using them to base their own predictions on, and setting the initial odds then it’s worth knowing this, even if you think it’s all voodoo.

I find it to be a fairly accurate reflection of most games I’ve watched, certainly more accurate than ‘shots’ or ‘possession’.

 

[Ginger Gnome]

In the Man City v Chelsea game the other week, XG was 4.04-0.90. There's absolutely no way that game was anything like a 4-1, especially when about .99 of one of those expected goals would be the easy one Aguero put wide. Like most stats it has some value, but doesn't tell the whole story. It's more useful for studying where teams' goals come from in general than individual matches (like everyone knows that on the Twitter updates on our games if it says "Corner to the Blades", the next Tweet will say "The corner comes to nothing" about 99% of the time).

These stats show that across the season Man City's expected goals are broadly in line with the actual number they've scored, although they've conceded less from corners than expected, but more from open play than expected. It also shows Jesus isn't putting his chances away.
https://understat.com/team/Manchester_City/2018
https://understat.com/team/Manchester_City/2018

 

[Revolution]

Found these from Blades Analytics tweets.

An xG table is interesting, but some people go further and say "this is what the table should be". That's going too far.

This stuff is one thing to add into the pot when teams are deciding who to buy, and who to play, and making tactical decisions. Many fans are also interested in it (I know I am: I picked up baseball sabermetrics from Bill James' writing 20 years ago. They are miles ahead in US sport). It is not the be all and end all. A well run club these days will have this data and look at it (along with other things not in the public domain) but there is more to management than running numbers. You have to deal with people, and that's the hardest part.

 

[FleaPitBlade]

It's all just one example of a generated number to measure and quantify something. How, when and why its interpreted deternines whether it's bullshit or useful.

I tend to trust my eyes where SUFC are concerned. I don't need a digit to tell me Norwood delivers a lot of key passes or that Billy Sharp is a clinical finisher. But the other teams and players in the division? I'm lucky if I've seen them half a dozen times on TV or twice in the flesh. That's why I like to understand the advanced stats and how they can add a bit of depth as a supporter - and, I suppose, it's the same from a scouting and tactical point of view for those in the game as well.

It's not just xG but passing and defending metrics - again, just numbers but they've been arrived at from reality and provide a glimpse.

 

[Eddiecolquhounsleg]

XG is the millennium equivalent of nasal strips from the nineties. They may look good and be mildly interesting but ultimately you had a plaster stuck on your fucking nose.

 

[FleaPitBlade]

Enda Stevens in a good example.

Look at the headline stats and last season his ''Assists" were quite low. People slated him for not creating enough and were calling for us to bring in competition.

The advanced stats like xG actually demonstrated he was creating good goal scoring chances but for whatever reason, more of what he created were missed by the finisher(s) than, say, those created by Fleck or Baldock. The advanced stats showed that his low 'assits' tally in relation to the rest of the team wasn't necessarily a fault in his play, more that it was an anomaly.

Roll that forward a 6 months and it averages back out and his assists contributions are more where you'd like them to be.

 

[PokerBlade]

Anything you observe about a football match can be quantified statistically. Whether it's passes made, distance ran, shots taken, or whatever else, you can stick a number to it and model it statistically.

So when someone says United don't cross early enough, you can in principle look at the number of crosses we make compared to other teams, number of passes in the build up, where possession was gained, position of the cross made, and position that most successful crosses are made from.

Of course there's a problem that this is extremely reductionist and football matches are probably quite chaotic. Changing one variable can have a huge impact on the overall team (if you decide to start crossing earlier, maybe you lose possession more, and maybe you concede more goals to the point that you're worse off even if scoring more). Overall though, stats are just a more detailed look at the kind of passing comment every poster on here makes. Whether stats are utilised well, that's another question.

 

[BushBlade]

Anything you observe about a football match can be quantified statistically. Whether it's passes made, distance ran, shots taken, or whatever else, you can stick a number to it and model it statistically.

So when someone says United don't cross early enough, you can in principle look at the number of crosses we make compared to other teams, number of passes in the build up, where possession was gained, position of the cross made, and position that most successful crosses are made from.

Of course there's a problem that this is extremely reductionist and football matches are probably quite chaotic. Changing one variable can have a huge impact on the overall team (if you decide to start crossing earlier, maybe you lose possession more, and maybe you concede more goals to the point that you're worse off even if scoring more). Overall though, stats are just a more detailed look at the kind of passing comment every poster on here makes. Whether stats are utilised well, that's another question.

I think it’s still in its infancy and, as you say, the nature of the game doesn’t lend itself to statistical analysis in the way a game like baseball does. There’s still a huge misuse of stats, particularly player stats for recruitment although I think clubs are wising up to this after the stupid strategies pursued by the likes of Villa. Stats are a tool and like any tool, have to be used correctly by people who know what they’re doing. That’s why we don’t give hammers to five year olds to play with.

 

[Carlton Blade]

Fuck me, is this what the beautiful game has been reduced to?

Depends how you look at it. I don't see it as a reduction and when I'm at the United match , or watching any football live I don't give it a 2nd thought.

But I enjoy analytics that help me figure out what a team is doing, why they might be doing it, what the other team do to counter and if any of it works post or pre match. I'm not a tactical genius and the stats help me get my head around things a bit better.

 

[Pinchy]

When they start awarding points for XG, then I'll start to worry about it.

Until then I'll watch the games and trust my own judgement as to which was the better team, who scored the most goals, and where the points are going just as I have done for the last 60+ yrs

Quite right too. I’m confident you’ll also avoid the most misapplied term in football, ‘lucky’.

 

[Pinchy]

Anything you observe about a football match can be quantified statistically. Whether it's passes made, distance ran, shots taken, or whatever else, you can stick a number to it and model it statistically.

So when someone says United don't cross early enough, you can in principle look at the number of crosses we make compared to other teams, number of passes in the build up, where possession was gained, position of the cross made, and position that most successful crosses are made from.

Of course there's a problem that this is extremely reductionist and football matches are probably quite chaotic. Changing one variable can have a huge impact on the overall team (if you decide to start crossing earlier, maybe you lose possession more, and maybe you concede more goals to the point that you're worse off even if scoring more). Overall though, stats are just a more detailed look at the kind of passing comment every poster on here makes. Whether stats are utilised well, that's another question.

Gerritinerlyyoonitedferfuckssaik.

 

[essexblade73]

The lads on the NTT20 podcast use xg as part of choosing their betting tips...and their win record is awful. Get the fckin ball on the pitch and play football.

 

[Chali 2na]

No, it isn’t. It’s absolute rubbish.

Thanks for your reply.

But what absolute bollocks that is. Think it’s the reply below yours gives the tap-in shot as a 0.99 XG. Fair enough. But say that’s Connor Sammon tapping it in. Is it then 0.09? Load of shit

It's not rubbish. It's just a stat, like any other.

You can choose to ignore it, like any other.

 

[PokerBlade]

I think it’s still in its infancy and, as you say, the nature of the game doesn’t lend itself to statistical analysis in the way a game like baseball does. There’s still a huge misuse of stats, particularly player stats for recruitment although I think clubs are wising up to this after the stupid strategies pursued by the likes of Villa. Stats are a tool and like any tool, have to be used correctly by people who know what they’re doing. That’s why we don’t give hammers to five year olds to play with.

There's a lecture by psychologist Robert Sapolsky called "Chaos and Reductionism" on youtube. Doubt many will be interested, but it runs through the problems of a reductionist approach that might well apply to a dynamic system like a football team. Basically, breaking things down to nice numbers works quite well most of the time, until it doesn't. At that point what you find is that breaking it down to constituent elements, hoping to eventually find the secret to how it works, turns into finding that slight tweaks to one part of the system has huge unpredictable results elsewhere. Weather systems are a good example. We can model things pretty well but at some point we start finding that for specific instances, it's all quite unpredictable.

Baseball is a pretty good example of a game that breaks down really well into stats. But it's not dynamic. It's stop-start, positions are well defined, there's a consistency to things like pitches. Football doesn't really work that way. You have 20 outfield players that are almost constantly changing position on the pitch, each individual performing different roles throughout the game and depending on the situation. Football's behind the rest of the world in how it utilises stats but I suspect there's always going to be some element that eludes us from finding the "perfect" football system.

 

[PokerBlade]

Quite right too. I’m confident you’ll also avoid the most misapplied term in football, ‘lucky’.

In poker, we prefer the term "variance".

 

[Pinchy]

There's a lecture by psychologist Robert Sapolsky called "Chaos and Reductionism" on youtube. Doubt many will be interested, but it runs through the problems of a reductionist approach that might well apply to a dynamic system like a football team. Basically, breaking things down to nice numbers works quite well most of the time, until it doesn't. At that point what you find is that breaking it down to constituent elements, hoping to eventually find the secret to how it works, turns into finding that slight tweaks to one part of the system has huge unpredictable results elsewhere. Weather systems are a good example. We can model things pretty well but at some point we start finding that for specific instances, it's all quite unpredictable.

Baseball is a pretty good example of a game that breaks down really well into stats. But it's not dynamic. It's stop-start, positions are well defined, there's a consistency to things like pitches. Football doesn't really work that way. You have 20 outfield players that are almost constantly changing position on the pitch, each individual performing different roles throughout the game and depending on the situation. Football's behind the rest of the world in how it utilises stats but I suspect there's always going to be some element that eludes us from finding the "perfect" football system.

POMO.

 

[Pinchy]

In poker, we prefer the term "variance".

I’ve had some fucking bad variance in my time...

 

[PokerBlade]

I’ve had some fucking bad variance in my time...

It might make you feel better to know that whatever you did the bad variance was always a statistically possible outcome outside of your control.

It might make you feel worse to know many of us won't live long enough to see that variance start to work in our favour.

 

[Pinchy]

It might make you feel better to know that whatever you did the bad variance was always a statistically possible outcome outside of your control.

It might make you feel worse to know many of us won't live long enough to see that variance start to work in our favour.

I once won over £100 in a “friendly” game of three card brag. It would have been a lot more but my opponent ran out of dosh...

He was holding three ladies...

 

[Ron_Justice]

I just take xG as a deeper version of the old 'that was a chance' v's 'that was a good chance'.

 

[SwindonBladeJnr]

There's a lecture by psychologist Robert Sapolsky called "Chaos and Reductionism" on youtube. Doubt many will be interested, but it runs through the problems of a reductionist approach that might well apply to a dynamic system like a football team. Basically, breaking things down to nice numbers works quite well most of the time, until it doesn't. At that point what you find is that breaking it down to constituent elements, hoping to eventually find the secret to how it works, turns into finding that slight tweaks to one part of the system has huge unpredictable results elsewhere. Weather systems are a good example. We can model things pretty well but at some point we start finding that for specific instances, it's all quite unpredictable.

Baseball is a pretty good example of a game that breaks down really well into stats. But it's not dynamic. It's stop-start, positions are well defined, there's a consistency to things like pitches. Football doesn't really work that way. You have 20 outfield players that are almost constantly changing position on the pitch, each individual performing different roles throughout the game and depending on the situation. Football's behind the rest of the world in how it utilises stats but I suspect there's always going to be some element that eludes us from finding the "perfect" football system.

Can't believe I've just read 'Robert Sapolsky' on a Blades forum!

 

[BushBlade]

It might make you feel better to know that whatever you did the bad variance was always a statistically possible outcome outside of your control.

It might make you feel worse to know many of us won't live long enough to see that variance start to work in our favour.

I occasionally watch the Poker World Championship and you often see someone betting hard on a very good hand with a probability of winning above 90%. Then the other player gets a river card to win. That’s a proper kick in the bollocks.

Football can be like that, you can play well, create sufficient chances to win and restrict the opponents to only a few half chances. But still lose 1-0.

 

[bricktop]

POMO.

There is a pretty good idea behind POMO, the thinking being that you can only score if the ball is in the ‘Position Of Maximum Opportunity), which makes perfect sense. Under the management of Weir and Clough I despaired as we fannied about with the ball and had no clear method of how to deliver the ball in to the POMO. To the untrained eye we might look like we fanny about with the ball too much but Wilder and Knill have trained the players several ways to quickly and effectively deliver the ball in to the POMO.

I personally think whoever came up with the term POMO was a coaching genius.

 

[BushBlade]

I once won over £100 in a “friendly” game of three card brag. It would have been a lot more but my opponent ran out of dosh...

He was holding three ladies...

I once cleaned a mate out playing 3 card brag, he even put his cigarettes in. I was begging him not to but he thought I was bluffing. I kept the bulk of it as I was a bit pissed off he thought I’d stitch him up like that, I wouldn’t do that to a mate.

 

[georgebernardshaw]

Football used to be a simple game, and trust me, I’m a simple man - maths used to be a nightmare for me, so much so that in a geometry O level exam, I drew an unlabelled triangle and then left for the day.

Assists in football used to be passes, referee’s assistants used to be linesmen, and any added time depended on the time of the ref’s train home - if we go All-American, the magic will have gone for me.