Uninsightful Statistics

[CasaStreets]

They seem to be everywhere these days. As I was reading the 'Rashford & Martial' thread I was reminded of how useless most of the statistics we quote are when taken in isolation... and thus, how narrow most commentary on this website is. It's either blindingly obvious or uselessly subjective.

It seems to me this is probably why "old school" scouts are increasingly supplemented and/or replaced with statisticians and data scientists who are creating new metrics to quantify player performance in the sport (and most high-value sports). Most Caf posters are really (really, really) poor versions of old school scouts - applying the "eye test" with some basic stats.

My question is, what content have you all found that shed more light on the statistics being used by the data science personnel at various clubs today? And are there any sources that make these types of datasets publicly available?

 

[acnumber9]

They seem to be everywhere these days. As I was reading the 'Rashford & Martial' thread I was reminded of how useless most of the statistics we quote are when taken in isolation... and thus, how narrow most commentary on this website is. It's either blindingly obvious or uselessly subjective.

It seems to me this is probably why "old school" scouts are increasingly supplemented and/or replaced with statisticians and data scientists who are creating new metrics to quantify player performance in the sport (and most high-value sports). Most Caf posters are really (really, really) poor versions of old school scouts - applying the "eye test" with some basic stats.

My question is, what content have you all found that shed more light on the statistics being used by the data science personnel at various clubs today? And are there any sources that make these types of datasets publicly available?

The problem with the ‘eye test’ is how subjective it is and prone to irrational bias. We’re not taking pass completion or ‘pressures’. We’re talking goals and assists. Describing them just as stats is fecking stupid.

 

[ClaytonBlackmoorLeftPeg]

The issue I see on here is the selective use of stats. The use of one metric that supports your argument/ perspective about a player.

however, let’s not get away from the fact this is a discussion forum, and football IS subjective.

Fans are biased, irrational and non logical.

You can have a favourite player who just isn't very good. You can think a young player is going to be great, when no one else sees it - and visa versa.

Stats will rationalise an opinion, and as the OP states enhance decision making.

I imagine clubs have a reasonable data set about their own players, use that as a benchmark to see where they can improve or enhance the squad - then use this as a basis to analyse the data collected by external agencies based on matches.

in terms of a public data set being freely available. Maybe, but surely one of the purposes of collecting such data is to monetise it and sell to clubs?

 

[Cast5]

I think stats are good in some situations for people who don’t understand what they’re watching or get told what to think of a player by a YouTuber or a Twitter account, like “x player misses the most chances” or “x player doesn’t score enough goals” you can tell them how many goals they’ve scored and their conversation rate compared to players who they’ve said are supposedly 100x better at taking chances.

I think they’re used poorly when people say Bruno’s pass completion rate is too low. You need to watch the game to see why it’s sometimes low, He’s obviously trying to create chances so makes high risk passes which is fine for your creative outlet that you wouldn’t want your centre half doing on the edge of his own box, this shouldn’t need to be explained to people who supposedly watch football regularly. Some people would like Gareth Barry at number 10 passing it back and sideways for 95% completion rate.

There’s a lot of stats “experts” who you talk to who actually know next to nothing about the game. It’s not necessarily stats that are a problem it’s how they’re used by certain people.

 

[roonster09]

The problem with the ‘eye test’ is how subjective it is and prone to irrational bias. We’re not taking pass completion or ‘pressures’. We’re talking goals and assists. Describing them just as stats is fecking stupid.

Exactly. CAF has turned into "I'm right no matter what", so they just twist everything to fit the agenda. When presented with stats, the reply is usual "Eye test" post.

 

[ClaytonBlackmoorLeftPeg]

Exactly. CAF has turned into "I'm right no matter what", so they just twist everything to fit the agenda. When presented with stats, the reply is usual "Eye test" post.

that happens, but also selective stats are often dug out, and used to “supplement” an agenda. Often used out of context, and other conflicting stats are ignored by the poster.

The challenge is that some people are trying to prove they are right, and others wrong. When football is largely about opinions and different perspectives.

 

[roonster09]

that happens, but also selective stats are often dug out, and used to “supplement” an agenda. Often used out of context, and other conflicting stats are ignored by the poster.

The challenge is that some people are trying to prove they are right, and others wrong. When football is largely about opinions and different perspectives.

Yeah agree with that too. People abusing "Pass completion %" "Dribbled past" "Possession loss" is jut awful for any discussion.

 

[ClaytonBlackmoorLeftPeg]

Yeah agree with that too. People abusing "Pass completion %" "Dribbled past" "Possession loss" is jut awful for any discussion.

One of the worst I see quoted is “interceptions per game”, when often the figures are below 1.

 

[Zen86]

The selective use of stats on here is just as subjective as the eye test. Football is a matter of opinions at the end of the day, and I’d much rather form mine based on what I actually see, than copying and pasting some numbers from a website.

 

[Hansi Fick]

I think one of the stats whose unreflective, "uninsightful" use bugs me regularly is the "shots on target" stat.
It is borderline meaningless when you consider that a shot that hits the post (and doesn't feature in the number) was a much more dangerous and relevant chance than a ball rolling straight at the goalkeeper, and when attackers evidently aim their shots for the fringe areas of the goal, not simply at goal.

Connected to that is the English commentators' staple "... but he's GOT to hit the target!"

 

[Andersons Dietician]

The truth is there are things in a game that can’t simply be measured by stats. So in reality they can be helpful but they never paint a full picture of what is actually happening.

We all have different opinions on how football should be played, what constitutes a good player and what value they bring to the team. Sometime people will have Bias towards players and see stats they can use to justify that players performance or position in the team.

Statman Dave is also just the worst.

 

[Pagh Wraith]

I think one of the stats whose unreflective, "uninsightful" use bugs me regularly is the "shots on target" stat.
It is borderline meaningless when you consider that a shot that hits the post (and doesn't feature in the number) was a much more dangerous and relevant chance than a ball rolling straight at the goalkeeper, and when attackers evidently aim their shots for the fringe areas of the goal, not simply at goal.

Connected to that is the English commentators' staple "... but he's GOT to hit the target!"

Over a medium-sized sample, shots on target is a very useful stat. It has the strongest correlation to goals scored (obviously) of any of the traditional stats and is a reasonably good predictor of future performances. Before xG came along, SoT was a staple for anyone in sports betting or predictive modelling or just to see which teams have been lucky or unlucky.

I think your post is more aiming at the usage in a single game and then I would agree. And with regards to the last bit, I used its little brother "got to score there" in the annoying football phrases thread. Completely meaningless statements.

 

[Pagh Wraith]

The truth is there are things in a game that can’t simply be measured by stats. So in reality they can be helpful but they never paint a full picture of what is actually happening.

We all have different opinions on how football should be played, what constitutes a good player and what value they bring to the team. Sometime people will have Bias towards players and see stats they can use to justify that players performance or position in the team.

Statman Dave is also just the worst.

I know what you are trying to say but that is a misleading statement and often used to discredit the use of data in football. If there is something that isn't captured by stats (and I totally believe there is) then it is because the stats and methods aren't sophisticated enough yet. Just think about how far we have come in the last few years. There used be only shots, corners and possession. Now we have pressures and key assists. It is an ongoing development and that is true for any science. Just because we cannot measure the probability of a striker missing an easy chance because he is affected by his girlfriend breaking up with him the night before, doesn't mean it doesn't exist. There is a number behind everything. Some are easy to obtain (such as counting the number of shots), some can only be estimated which is why we use simplified models.

I do also believe though, that we are at a point now where very little slips though the radar in terms of performance analysis. Doesn't mean human aren't capable of widely misapplying data.

 

[VorZakone]

Didn't Liverpool sign Keita after lots of statistical analysis? Turned out to be an underwhelming signing.

I think Arsenal also had a stat agency advising them on signings which turned out to be bang average.

 

[GoldTrafford99]

Most stats are bullshit; expect this one.

Any one player only ever has the ball at their feet for - on average - 1.8% of any game.

That means 98.2% of a match for any player is played off the ball.

Yet current fans NEVER rank players' involvement off the ball. All modern fans are interested in is possession and goals and assists and pass accuracy stats. But sure, that's limiting a players involvement in the game to just the 1.8% of the time they use the ball over 90-minutes.

The best players - certainly the most underrated ones - such as Busquets, Carrick, Henderson etc - all of their best work during any 90-minutes is carried out in the 98.2% of the time that 'fans' don't even get to see...

May I also add, Marcus Rashford off the ball is absolutely sublime (one of the best I have ever seen)... yet, all y'all in the Rashford Thread don't see the amazing runs he makes off the ball to get into the positions he gets into... all you see is the fact that the shot he managed to get off (which most other footballers wouldn't have got off because they simply wouldn't have been as effective off the ball) didn't hit the net, and therefore he is shit.

 

[Glorio]

Didn't Liverpool sign Keita after lots of statistical analysis? Turned out to be an underwhelming signing.

I think Arsenal also had a stat agency advising them on signings which turned out to be bang average.

I think you'll find most top teams sign players after lots of statistical analysis and have a data science team affiliated with them

 

[Classical Mechanic]

They seem to be everywhere these days. As I was reading the 'Rashford & Martial' thread I was reminded of how useless most of the statistics we quote are when taken in isolation... and thus, how narrow most commentary on this website is. It's either blindingly obvious or uselessly subjective.

It seems to me this is probably why "old school" scouts are increasingly supplemented and/or replaced with statisticians and data scientists who are creating new metrics to quantify player performance in the sport (and most high-value sports). Most Caf posters are really (really, really) poor versions of old school scouts - applying the "eye test" with some basic stats.

My question is, what content have you all found that shed more light on the statistics being used by the data science personnel at various clubs today? And are there any sources that make these types of datasets publicly available?

I think this is the best free resource.

https://fbref.com/en/players/a1d5bd30/Marcus-Rashford

It compares players against every other player in Europe’s top 5 leagues over numerous metrics.

Didn't Liverpool sign Keita after lots of statistical analysis? Turned out to be an underwhelming signing.

I think Arsenal also had a stat agency advising them on signings which turned out to be bang average.

That’s mentality. It’s much more intangible. It’s hard to assess if a player will shrink in a stronger field, under the weight of a famous shirt, the pressure to win every week and more exposure. Clubs do scout personalities as well but it’s harder to pin down.

 

[Theo Cherry]

The truth is there are things in a game that can’t simply be measured by stats. So in reality they can be helpful but they never paint a full picture of what is actually happening.

We all have different opinions on how football should be played, what constitutes a good player and what value they bring to the team. Sometime people will have Bias towards players and see stats they can use to justify that players performance or position in the team.

Statman Dave is also just the worst.

Like what?

 

[dumbo]

Uh.. yeah definitely.

 

[Classical Mechanic]

Uh.. yeah definitely.

I had one of your posts from another thread it my quote list.

 

[dumbo]

I had one of your posts from another thread it my quote list.

Yeah I assumed so. But now I'm curious/paranoid.

 

[Classical Mechanic]

Yeah I assumed so. But now I'm curious/paranoid.

I generally browse the CAF on an iPad so I must have hit quote by accident scrolling down. I wasn’t intending to reply to the post.

 

[The Original]

Most stats are bullshit; expect this one.

Any one player only ever has the ball at their feet for - on average - 1.8% of any game.

That means 98.2% of a match for any player is played off the ball.

Yet current fans NEVER rank players' involvement off the ball. All modern fans are interested in is possession and goals and assists and pass accuracy stats. But sure, that's limiting a players involvement in the game to just the 1.8% of the time they use the ball over 90-minutes.

The best players - certainly the most underrated ones - such as Busquets, Carrick, Henderson etc - all of their best work during any 90-minutes is carried out in the 98.2% of the time that 'fans' don't even get to see...

May I also add, Marcus Rashford off the ball is absolutely sublime (one of the best I have ever seen)... yet, all y'all in the Rashford Thread don't see the amazing runs he makes off the ball to get into the positions he gets into... all you see is the fact that the shot he managed to get off (which most other footballers wouldn't have got off because they simply wouldn't have been as effective off the ball) didn't hit the net, and therefore he is shit.

Well following the Pareto principle, clearly what happens on the ball is much more relevant to the outcome of the game. At the bottom line if an attacker is getting into good positions but not scoring how useful is he?

 

[GoldTrafford99]

Well following the Pareto principle, clearly what happens on the ball is much more relevant to the outcome of the game. At the bottom line if an attacker is getting into good positions but not scoring how useful is he?

Well extremely useful if he is causing trouble for the opposition... whether he is scoring or not. Somebody else will score eventually from the space provided. How do we think Emile Heskey got 60-odd caps for England even though the guy couldn't hit a barn door with a guitar? Michael Owen says Heskey is the best strike partner he ever played with... now, just think about all the strike partners Owen had in his career from Ronaldo to Fowler to Berbatov...

Obviously the dream striker would be great off the ball as well as with his finishing; no better player in my opinion for this than Alan Shearer.. some reader of the game and some finisher (the best I've ever seen). Marcus Rashford is also awesome at it... Cavani's brill too. Our manager was one of the best at it too... Their best skills are off the ball... not on the ball...

The Pareto Principal doesn't apply in football... simply because that's suggesting the work being done off the ball is insignificant, when in fact it is just as pivotal - even moreso - as the work being done on the ball...

 

[Andersons Dietician]

Like what?

Can you determine if the sun was in someone’s eyes and at what brightness? Can you determine and measure the spin on a ball, or a bobble in a surface, how a strong breeze effects the movement or speed of a ball, did the player passing the ball to you put you in trouble, was it knee high, shin high, thigh high, did a cat run on the pitch and deflect the ball, did someone do something that caught a players eyeliner in the crowd. Did the gel in the players hair gel wet and drip in to his eyes.

There are so many variables that anyone who thinks stats and stats alone are all you need and actually observing a game means nothing I don’t have any time for you.

Heck there is a whole film built off of this very subject. Money Ball.

 

[The Original]

Well extremely useful if he is causing trouble for the opposition... whether he is scoring or not. Somebody else will score eventually from the space provided. How do we think Emile Heskey got 60-odd caps for England even though the guy couldn't hit a barn door with a guitar? Michael Owen says Heskey is the best strike partner he ever played with... now, just think about all the strike partners Owen had in his career from Ronaldo to Fowler to Berbatov...

Obviously the dream striker would be great off the ball as well as with his finishing; no better player in my opinion for this than Alan Shearer.. some reader of the game and some finisher (the best I've ever seen). Marcus Rashford is also awesome at it... Cavani's brill too. Our manager was one of the best at it too... Their best skills are off the ball... not on the ball...

The Pareto Principal doesn't apply in football... simply because that's suggesting the work being done off the ball is insignificant, when in fact it is just as pivotal - even moreso - as the work being done on the ball...

The Pareto Principle applies to all human endeavour-this is well studied and well-documented. To argue against this is akin to arguing against gravity. More so when it is particularly easy to prove that the principle particularly applies to football. Simply, the objective of football is to win games. The output of the game is winning, losing, or drawing. In any such event, the principle simply states that 20% or less of everything that you did was the major cause of the outcome. This is demonstrably true. While the game is played over 90 minutes, the passage of play that will lead to a goal, rarely exceeds 2 minutes a most.

The Pareto Principle also doesn't mean that the rest of your inputs into the game do not matter at all, they are simply less important. In that sense, a footballer's most important contribution is with respect to what they contribute on the ball, with their off-ball contributions coming in, still important, but secondary.

On Rashford and the players you compare him two, you are mixing two distinct arguments. One, that a player can be good off the ball and less good on the ball without being a bad player, and two that a striker can contribute in other ways without being a great scorer. This is extremely distinct, because the strikers you referenced (at least Heskey), and indeed, other players of that mould, contributed by being target men, knocking the ball down for their smaller, quicker partners, such as Owen. That is on-the-ball work, not off the ball. Crucially, Rashford is not even that mould of a player, and his on the ball contributions to other players is average.

Further, and separately, his off the ball work scarcely yields opportunities for other players and while it creates chances for himself, he fails to convert efficiently, quite unlike the players you mentioned, and this is a crucial distinction. Thus, if his off-ball contribution leads to no end product, there is little ground for arguing that he is a good player because of his off-ball contribution.

In exceptional cases, a footballer is more crucial for his off the ball work, but these will usually be very limited players. Ji Sung Park for example, or Dirk Kuyt. Players, who, while respected for their off the ball attributes, will rightly be remembered as being below average when compared to the top players in their position.

 

[Needham]

Can you determine if the sun was in someone’s eyes and at what brightness? Can you determine and measure the spin on a ball, or a bobble in a surface, how a strong breeze effects the movement or speed of a ball, did the player passing the ball to you put you in trouble, was it knee high, shin high, thigh high, did a cat run on the pitch and deflect the ball, did someone do something that caught a players eyeliner in the crowd. Did the gel in the players hair gel wet and drip in to his eyes.

Can a person pose so many rhetorical questions without question marks.
Seriously though put that to music, I'd download it. You're right and all, if football was entirely reducible to numbers it would be a science and not the art we love it for being.

 

[Hammondo]

Over a medium-sized sample, shots on target is a very useful stat. It has the strongest correlation to goals scored (obviously) of any of the traditional stats and is a reasonably good predictor of future performances. Before xG came along, SoT was a staple for anyone in sports betting or predictive modelling or just to see which teams have been lucky or unlucky.

I think your post is more aiming at the usage in a single game and then I would agree. And with regards to the last bit, I used its little brother "got to score there" in the annoying football phrases thread. Completely meaningless statements.

This is s really typical response to this type of thread.

Your way of using a stat is bad, but mine is good. It's frustrating when you don't seem to take in what's being said.

Stats alone are useless, you need full context of what you are comparing to what. You need lots of data and stats to create the context and backup a fleshed out idea.

One stat, like shots on target tells you feck all or basically nothing.

 

[Hammondo]

I think you'll find most top teams sign players after lots of statistical analysis and have a data science team affiliated with them

Yea but they don't make decisions just from statistics. Those statistics will be the start of the process.

 

[passing-wind]

Football is and always will be a game formed of opinions especially in a speculative manner. One thing's for certain however many posters especially in the match day thread have some really odd views about the positives to be taken out of team performances which I will never understand. You'll only have to go to the Maguire thread to see suggestions that he's one of the best defenders around to highlighting the extent of subjectivity when we all see players perform.

 

[passing-wind]

Yea but they don't make decisions just from statistics. Those statistics will be the start of the process.

Absolutely there are some metrics that stats don't provide any accountability for especially when contrasting the differences in one league playing environment to another. Not to mention tactical accomodation, if player X plays well in a particular system can that success be replicated in another team that otherwise has a completely different position available as an equivalent. Those scenarios are in need of judgements more than anything else.

 

[Andersonson]

Most stats are bullshit; expect this one.

Any one player only ever has the ball at their feet for - on average - 1.8% of any game.

That means 98.2% of a match for any player is played off the ball.

Yet current fans NEVER rank players' involvement off the ball. All modern fans are interested in is possession and goals and assists and pass accuracy stats. But sure, that's limiting a players involvement in the game to just the 1.8% of the time they use the ball over 90-minutes.

The best players - certainly the most underrated ones - such as Busquets, Carrick, Henderson etc - all of their best work during any 90-minutes is carried out in the 98.2% of the time that 'fans' don't even get to see...

May I also add, Marcus Rashford off the ball is absolutely sublime (one of the best I have ever seen)... yet, all y'all in the Rashford Thread don't see the amazing runs he makes off the ball to get into the positions he gets into... all you see is the fact that the shot he managed to get off (which most other footballers wouldn't have got off because they simply wouldn't have been as effective off the ball) didn't hit the net, and therefore he is shit.

Great post. 100% agree with this

 

[Hansi Fick]

Over a medium-sized sample, shots on target is a very useful stat. It has the strongest correlation to goals scored (obviously) of any of the traditional stats and is a reasonably good predictor of future performances. Before xG came along, SoT was a staple for anyone in sports betting or predictive modelling or just to see which teams have been lucky or unlucky.

I think your post is more aiming at the usage in a single game and then I would agree. And with regards to the last bit, I used its little brother "got to score there" in the annoying football phrases thread. Completely meaningless statements.

Yes. In any case, the bigger the sample size, the better stats become, and a football game is an event of a very small and unsteady sample size of stuff that happens.

 

[Borys]

This is s really typical response to this type of thread.

Your way of using a stat is bad, but mine is good. It's frustrating when you don't seem to take in what's being said.

Stats alone are useless, you need full context of what you are comparing to what. You need lots of data and stats to create the context and backup a fleshed out idea.

One stat, like shots on target tells you feck all or basically nothing.

That's because every now and then some not-so-smart people moan about use of statistics and give individual incidents as an argument against it ("it's all bullshit"). I think @Pagh Wraith explained it nicely, or is there any specific point you disagree on?

Most stats are bullshit; expect this one.

Any one player only ever has the ball at their feet for - on average - 1.8% of any game.

That means 98.2% of a match for any player is played off the ball.

Yet current fans NEVER rank players' involvement off the ball. All modern fans are interested in is possession and goals and assists and pass accuracy stats. But sure, that's limiting a players involvement in the game to just the 1.8% of the time they use the ball over 90-minutes.

The best players - certainly the most underrated ones - such as Busquets, Carrick, Henderson etc - all of their best work during any 90-minutes is carried out in the 98.2% of the time that 'fans' don't even get to see...

May I also add, Marcus Rashford off the ball is absolutely sublime (one of the best I have ever seen)... yet, all y'all in the Rashford Thread don't see the amazing runs he makes off the ball to get into the positions he gets into... all you see is the fact that the shot he managed to get off (which most other footballers wouldn't have got off because they simply wouldn't have been as effective off the ball) didn't hit the net, and therefore he is shit.

Best example of that is how teams are organized to press. It's hard to measure. But this applies to a team, and stats usually refer to players performance in detail.

 

[UpWithRivers]

Its not just football fans, its the pundits, journalists and even the professionals. Thats the beauty and the bane of football. I saw Neville using stats to prove that Maguire and Lindelof arnt as good as Vidic and Stam or Bruce and Pallister. But how can you ever quantify those stats when each pair played in different teams altogether. At the end of the day its obvious to everyone they arnt better but saying they let in less goals per game proves nothing.

 

[Theo Cherry]

Can you determine if the sun was in someone’s eyes and at what brightness? Can you determine and measure the spin on a ball, or a bobble in a surface, how a strong breeze effects the movement or speed of a ball, did the player passing the ball to you put you in trouble, was it knee high, shin high, thigh high, did a cat run on the pitch and deflect the ball, did someone do something that caught a players eyeliner in the crowd. Did the gel in the players hair gel wet and drip in to his eyes.

There are so many variables that anyone who thinks stats and stats alone are all you need and actually observing a game means nothing I don’t have any time for you.

Heck there is a whole film built off of this very subject. Money Ball.

Stats are there to paint a picture, all those isolated/incidental things aren't that important in the grand scheme of a game.

 

[Hammondo]

That's because every now and then some not-so-smart people moan about use of statistics and give individual incidents as an argument against it ("it's all bullshit"). I think @Pagh Wraith explained it nicely, or is there any specific point you disagree on?


Best example of that is how teams are organized to press. It's hard to measure. But this applies to a team, and stats usually refer to players performance in detail.

Well he says 2 things from what I can see. 1 is that shots on target correlates well with goals scored, no shit. No context so it actually tells you basically nothing. Shots on target tells you one thing, shots on target. It can only suggest at other things and without a huge amount of data it cannot tell you anything else.

This is why there is such a big industry built around it using immense computing power and a lot of expertise, because the amount of data needed for context is huge. Picking one stat and pointing at it is a caveman understanding of statistics.

The second thing he talks about is that if we cannot get the data then its because our means is not developed enough, which is true, but will be true for our lives at the least. If we could get a good picture of things with the data we can collect then coaches and managers would not be needed, instead the data is far behind them.

 

[Borys]

Well he says 2 things from what I can see. 1 is that shots on target correlates well with goals scored, no shit. No context so it actually tells you basically nothing. Shots on target tells you one thing, shots on target. It can only suggest at other things and without a huge amount of data it cannot tell you anything else.

This is why there is such a big industry built around it using immense computing power and a lot of expertise, because the amount of data needed for context is huge. Picking one stat and pointing at it is a caveman understanding of statistics.

The second thing he talks about is that if we cannot get the data then its because our means is not developed enough, which is true, but will be true for our lives at the least. If we could get a good picture of things with the data we can collect then coaches and managers would not be needed, instead the data is far behind them.

I don't understand what you're arguing about. You sound very patronizing with your "such a typical response/your way is bad mine is good" yet you don't give any specific examples of "context".

I think we basically agree on all points, there is no argument context is very important and you need to analyze big sets of data. This is why you actually need people analyzing the available data, unless the cause and effect relationships are well understood but luck plays a big role in football anyway.

 

[giorno]

Didn't Liverpool sign Keita after lots of statistical analysis? Turned out to be an underwhelming signing.

I think Arsenal also had a stat agency advising them on signings which turned out to be bang average.

And Salah. And Mane. Robertson, Van Dijk, Jota, etc. Every player they've signed under FSG was the result of lots of statistical analysis. Which got increasingly better as they improved their models and thus got better information out of it

Arsenal afaik didn't actually listen to said company's advice most of the time. Still, stats are only as good as the people who compile them and those who read and interpret them. Barcelona have one of the best departments of statistical analysis in football. Too bad they only use them as a PR stunt instead of actually using it their data

 

[Chairman Steve]

Ebbe Skovdahl, manager of at the time Aberdeen once said “Statistics are like miniskirts. They give you good ideas but hide the most important things”

Stats can only give you so much until you can actually watch a player in person to judge them in my opinion.