Do Arsenal fans rely on unreliable and misleading ‘stats’ too much nowadays?

The over-reliance on spurious stats  by Jon Fox

My fellow Gooners, for ages I have been depressed at how so many fans have such a slavish devotion to stats and wrongly assume they are reliable guides, when plainly, in many cases, they are not.

In truth, I have long been of the firm opinion that a slavish devotion to the use of statistics (stats) as a reliable guide to a player’s ability is based on a fallacy.

I do not say, of course not, that all stats are useless. That would be silly and also, wrong. But to rely on stats as being reliable   information to a player’s ability is way overblown by many fans in general and  Gooners seem no exception, if one takes takes as truth the many posts which use them slavishly and without intelligent context.

I am the first to agree that when used in context as a guide, though not as an infallible one, that many stats are indeed useful. But my purpose in this article is to try to point out to some who seem not to ever think beyond first base, so to speak, how misleading SOME stats can be and indeed, often are.

In common with many fans, I look closely at the now common use of stats in newspapers and online, though in my case I have an ingrained cynicism towards them always being of limited real use.

As an older fan I easily remember the times when the only stats published in sports sections of papers were the score, the scorers, time scored and attendance.  Oh, plus the half time scores.  At odd times through the season a list of leading goal scorers might well appear, but this was  random at best, and “assists” were never even thought about.

As the years and decades have passed, there has been more and more, relentlessly so, concentration on the minutiae surrounding any top-level game and this has in turn spawned a whole stat based industry. The use of computers has of course intensely speeded up this creeping intrusion into what is really a rather simple game.  Not that one would think it’s simple upon reading all the increasing waffle and trivia that surrounds any game.

I have, I confess, been wanting to write this debunking of stats   piece ever since the day when Mertesacker and Koscielny played together as our CB pairing.  I would regularly read that Merts was among the Prem’s most reliable and passers, overall. That was of course nonsense, but the stats said he was!

What those stats did not say was that almost all those passes, which gave him an approx. 95% successful pass rate, were made when under no opponent pressure and he and Kos would pass the ball back and forth, again and again (while I grew a beard just watching this boredom!!!) til Kingdom come. But that “stat” put him firmly among the Prem’s very best one or two passers.

My fellow Gooners I DO put some credence in stats which give a striker an overall minute played per goal, or goal and assist. Jesus comes out rather well, even in his City days, when you look into that stat.

I consider it a good guide – provided it is used only on players who have played a seriously decent number of games, and not on those who have played only a tiny few. There is some safely in use over a long time, but if in short bursts only, there are bound to be many fallacies. It is about using one’s brain to closely consider what use, if any, and if any, then how much, use do all stats have. IF fans can do that and many can and do, then stats, used wisely are helpful, though not as sole arbiters of a players use.

It is widely known that top clubs, in particular, do use detailed stats for players who they may be considering buying.  I have no beef with that, because within most clubs there rare wise people in charge of how stats are used, and they know how much reliance to place on them and how much not to place.

But it is we fans who tend to use stats in amateurish ways, then project our own opinions on to their use as if to back up our opinions. By all means have opinions; we would all be up a gum tree without opinions.  Opinions are the life blood of fanhood, and I love seeing, hearing and reading them. Though we all have a responsibility to use our brains wisely and not to lazily trot out false stats to “back up” our opinions. THAT is the whole purpose of this piece; to make us think before we trot out nonsense, based upon incorrect use of stats.

IF I can convince even one person not to continue doing what he/she has done previously, then this article will not have been written in vain.

Thank you for reading this and please do at least consider whether or not YOU personally are a wise or a lazy user of stats. Used in context, they DO have real role to play… but not otherwise.

COYG!

Jon Fox

Tags arsenal stats

31 Comments

  1. Arsenal had six relatively-disappointing players, who didn’t live up to their StatDNA scouting system results. They’re Calum Chambers, Lukas Podolski, Mathieu Debuchy, David Ospina and Olivier Giroud

    On the contrary, Liverpool signed several excellent players using their data analytics scouting system, such as Salah, Mane, Robertson and Wijnaldum. They also made some mistakes, such as the injury-prone Naby Keita, but I believe Arsenal made more wrong deals

    I think we should consider all stats of a player, rather than just the flashy ones such as goals and assists. I bet the big clubs’ data analytics scouting systems don’t calculate a player’s injury history and his compatibilities to various systems

    1. I might agree with you but your examples are quite flawed.

      De buchy was injured yet you fail to mention it as you do for Ketia.

      OG was not a bad buy and even went to Chelsea and won something.

      1. I actually quoted Daily Star’s article for the examples 😛

        I forgot to mention the sixth example, Mesut Ozil

        1. But Ozik when he arrived was fantastic and led tonus winning silverware he also almost broke the assist record.

          Yes he fell away and became a liability, but the first 3 years or so he was performing well so I would say the data analysis was pretty good.

  2. You are preaching to the converted when it comes to me Jon.Basically computer wizardry has been used, at considerable expense, to add another ,highly paid layer , to the cost of running a Football Club which hitherto was a community based business with which fans could identify and formed a major part of their recreational life.In the case of certain fans who are priced out of actually attending matches or unable to do so for other reasons, stats have become an integral part of their outlook towards football.They form a major part of their assessment of players when in fact they can be completely misleading as in the case of me running twenty yards, beating 3 defenders then passing to Jon Fox who lays on a one yard, sideways pass for Sue P to score with a tap in.Sue gets the glory, Jon gets the assist and poor me who made it all happen gets what exactly?To be fair stats are extremely helpful when it comes to analysing the fitness record of players which does form a major part of the due diligence procedure so necessary in view of the huge transfer fees paid today.As far as scouting is concerned eyes and intelligence beats stats anytime.

    1. Grandad, for good measure, you’re gon get carries into the final third, and maybe pre assist. Heard they’re integrating that as a stat too these days

    2. “Basically computer wizardry has been used, at considerable expense, to add another ,highly paid layer , to the cost of running a Football Club”.

      This could not be more incorrect and actually adds considerable business value while decreasing expenditures in scouting.

      The benefit far out weight cons and business analytics is a highly valued tool when done correctly.

    3. @Grandad, when I saw your post something came to mind. Remember the opening goal against Nottingham Forest last week? Did you see who saved the ball and passed to Martinelli before it went to Saka and back to Martinelli? Takehiro Tomiyasu.

      Yes, the same Tomiyasu that is being maligned as not better than Tierney. When people remember how they felt when Arsenal were only up by THAT one goal…..you see where I am going with this? I am not typing this to attack anyone, but it serves to buttress how much I have admire Mikel Arteta for thinking outside the box. He sees things in training we do not – and when his plans work, they work really well. Thanks for indulging me 🙂

  3. Jon, you actually make a great point about it being overused by fans in general. Ive never been big on quoting stats in all honesty, but at the same time it’s an important aspect of the game.
    A lot of stats are useless though, I don’t know about others but the only stats I find important when trying to talk about a player or team are the goals, assist, and chances created, and interceptions. I find the rest useless, because most of them can be misleading.
    That’s not to say goals and assists aren’t sometimes misleading too, as players can have poor games and just score a goal and fans would hype them up.
    Look no further than Benzema’s comment regarding stats.

    Fans overuse it too much, but the truth is you can’t do without it also. When discussing football and analyzing games, you can’t work on eye tests alone, and you can’t work on stats alone. It actually goes both ways, you need a bit of both to actually set things right because the numbers back up the eye tests and what you see.
    Take for example, players that go under the radar in different teams despite performing too well, when you tend to go back to see the games and see their contributions through numbers, you’ll know the performance is actually there too.
    Take the likes of Diogo Dalot this season, Ben White, Bernando Silva and couple of others. You’ll see how important they are when you see their contributions goes under the radar, meanwhile you’ll see the other guys who make the final play that’s to goal get all the plaudits just like Grandad pointed out.
    So really modern football actually works better with the eye test alongside the stats.
    The use of it needs to be done consciously and not get overblown

    1. Another example was Martinelli before Arteta, Martinelli of 2 years ago who used to look very good at doing nothing but running around headlessly and misusing his energy and would get a couple of goals.
      The eye test merchants alone would argue with you all day that he was this, he was that and yada yada yada when in all truth the whole running around headlessly convinces you he was doing great when in all honesty he doesn’t actually influence the game a lot except what fans called his energy.

      Now compare Martinelli from 2-3 years ago to the Martinelli we have right now.
      Since he was asked to channel vhis energy more and to stop doing everything at 100km p/h, eye tests will tell you he’s changed immensely and influences game now, and the stats are there to back him up. If you see his figures now in his all round play, you’ll see the difference. His energy, eye test, stats and everything backs him up.
      That’s why both are actually a good part of the game. They just need to be used properly.

      1. So an 18 year old with very little tutelage but with undoubted talent was “running around headlessly”
        My friend, you are a buffoon.
        Martinelli, in fact stalled mostly because Arteta preferred willian. It was when arteta decided to change approach to ‘project’ when everything started to gel. Martinelli was young raw but better than almost all out attackers- just go and watch his first season. He is still full of running but now, he clearly has a defined role in the team. Unlike before

  4. I do think that they have developed some more useful stats in recent years – presumably to help clubs with scouting. For example, the xG value does have some value in showing how good a player is at getting opportunities in a given team, and comparing with actual goals scored tells you how good a finisher they are to an extent. The way they determine xG does seem to be very complicated and I wouldn’t assume it to be a perfect measure, however.

    In any case, I’d always judge based on what I see first and foremost. Stats will always be secondary to me.

      1. Thanks Davi and I see you mention xG. Though I have never understood quite how it is calculated. So have you any idea that would help me, please?

        What I mean is who is it “expected” by and what is that “expectation” based on? It seems likely to be rather subjective to me , unless there are reliable and fair to all methods of calculation. Perhaps there indeed are, but how do we know?

  5. There are some people that have been burying themselves in this “Stats” overindulgence for far too long.

    You sometimes get to a point where you might as well be interacting with a computer/ robot..

    You try to go with what you saw with your eyes but the Stat monsters will tell you that these computers generated opinion are more reliable than my own eyes.. For example you watch a player having a complete nightmare of a game but at the end of the game the computer tries to assure you other side. He completed the most passes, he made the most tackles and headers. Most chances created blah blah..
    So the computer generated out of context Stats become template. Your eyes are just biased.

    Me:
    Player “A” had such a poor game.

    State monster:
    “But the computer generated stats says the opposite, it clearly shows that player “A” had a fantastic game. Look at his distance covered, look at his chances created, look at his interception stat, look at his Breathe per minute rate stat etc, they all show that he had a very good game. Stop being biased against player “A”. Lol

    The only stat’s that I can take seriously are those that monitor health / fitness..
    And even those ones should be taken with a grain of salt.

    All Stats should be taken in context. Can’t just be spewing out Stats and not including the context.

  6. Statistical systems are far more advanced these days than those displayed on public-facing websites like WhoScored. Generally, even the measured variables are not created by humans but are often compiled using machine learning algorithms to determine what aspects of unlabeled data most impact things such as win-share over expected, etc. This trend started off in baseball and has now made it to soccer. Of course, the eye test at the end of the day is still necessary to verify the authenticity of these stats, as is a decently large sample size. Nonetheless, I would never criticize management or a scouting team to use all relevant data to identify players that could be a good fit.

  7. Good article Jon, as the saying goes, “ lies, damned lies and statistics “. I think the whole idea of stats influence came from American football and baseball,

  8. The most important “stat” for any football supporter, is what they see with their own eyes.
    That’s why I feel very privileged to have a season ticket to watch our games as they happen, rather than through the lenses of a camera, that has to follow the ball.
    The after match summary is a very good way to discuss these “off the ball” incidents and I thoroughly enjoy watching them.

    Now Jon has mentioned Mertesaker in his article, a player he has always derided as being too slow etc etc.
    Now, that’s an opinion that doesn’t hold up when one compares the stats of his career in football, both at club and international level and one can use stats to counteract his views for a healthy discussion.

    When watching a game, one can see players interpassing, with no assumed benefits, unless one can see the panoramic view of players running into spaces, in order to receive the ball.

    Opinions are down to individual people during a game – stats provide what actually happened during a game… take them with a pinch of salt is my recommendation.

    Now, those like Jon, might find some of those stats irrelevant and I think most intelligent football fans will use their eyes when appraising a players performance.

    But here’s a classic example for us to ponder :
    Gabriel stats show that he hasn’t scored in ex number of games and fans are starting to question him.
    However, if one looks at the workrate, creative work with the ball, creative work off the ball, yards covered… then one gets the complete picture of just what a player he really is… and that’s where stats become invaluable to, say, MA.

    Surely, it’s an individuals choice to use stats, when discussing a point, or appraising a performance overall, or even to counteract a fellow fans opinion?…. Mertesaker being a case in point!!

    In summary, stats can be used to help an opinion – likewise they can be used to hinder said opinion.
    Most supporters use their eyes to judge a players performance.

    1. KEN, SKILLFULLY AVOIDED MY SPECIFIC POINT ABOUT MERTS though! I saw no need in an article about stats, to mention his lack of running speed. But I note that you did.

      I never used stats to determine his speed(or lack of) as there was no need, eyes being perfectly adequate.

      Do you OR do you not think that Merts was among the most consistently successful Prem passers though, as the misleading stats would have us believe? I never thought he was, not remotely so.

      Your last two sentences back up my whole piece, extremely well BTW! So (apart from your needless dig at me about me saying Merts was too slow) did your whole post.

      1. Jon, are you know denying that you have said, on many occasions, that Mertesaker was slow?
        I don’t understand why you change your mind so much, but that’s not a problem.

        As for Merts being a consistentently good passer of the ball, I have no doubt at all that he was…. as the stats over his whole career prove, not just the PL… at both international and club level.

        Successful football players aren’t so, if they can’t pass a ball properly, just ask ALL those German World Cup winners.

        I know I proved your article (apart from Mertesaker) correct, why would you think otherwise?

        1. Ken, Of course I DONT retract, even an iota, that Merts was, IMO, the slowestplayer who ever wore our shirt.

          Whatever gave you the idea that I RETRACTED IT?

          I SIMPLY SAID IT WAS POINTLESS AND NEEDLESS TO MENTION IT, NOT THAT IT WAS NOT TRUE.

  9. Jon…

    It was a pleasure to read your beautifully written and constructed piece! Statistics should be consumed with intelligence, and I agree wholeheartedly!

    However, as with nearly every aspect of life, it is said that variety is the spice of life. Similar to how we visit the mall and admire the seemingly endless variety of products on display, we only purchase or consume those items that appeal to us.

    Similarly, I believe the statistical industry has grown to provide a vast repository of statistical variables, and customers now have access to a vast array of statistical products.

    Consequently, I would rather criticize the production of these statistics, arguing that statistical measures ought to be developed with intelligence. However, since we have no control over who and how these measures are developed and reviewed, we can only advocate for their consumption with some level of critical thinking.

    Again, thank you for this thought-provoking call!

    Remain blessed!

    Cheers!

  10. 😂😂😂…

    This has kept me laughing for quite some time!

    Thank you, Jon, for brightening my day with your fantastic sense of humor!

    Cheers!

  11. I’d agree that “stats” are used badly by many people.

    But – there’s a world of difference between some football fan spouting a random number to justify their opinion and statistics, the branch of mathematics.

    I don’t like to do what’s called “arguing from a position of authority” (saying things like “I have a degree in statistics, so I must be right” or even “trust me, I’m a doctor”), but in thsi case there’s probably no alternative. So please excuse me on that front in advance…

    Statistics, mathematically, consists of a number of tests and an approach. Each of those tests has a purpose, often it’s to test a hypothesis which is often to determine whether a predictor variable has a statistically significant relationship with an outcome variable.

    So if you want to know if the colour of a team’s socks (the predictor) has a signifiant effect on the outcome of a match ( the outcome variable), you’d gather data on matches won and lost by teams wearing red socks and probably apply what’s known as a regression test (a piece of mathematics designed to test this hypothesis).

    More likely, you want to know something less frivolous.

    A good example was the Oakland Athletic baseball manage Billy Beane, who employed a statistician to test which predictors mattered most. He had to operate on a low budget compared to other teams and realised that the old way of having “experts” (often scouts) argue subjectively over who to sign wasn’t going to do anything to help him solve his money problem.

    The guy he hired found something surprising – that there two most important predictors were not the ones people expected, as stolen bases, runs batted in, and batting average.

    Properly-applied statistical tests / analysis showed that on-base percentage and slugging percentage are better indicators – at least of *offensive* success.

    So they set about buying players who were highest in those variables – with remarkable success. You can see the story told in the book or film Moneyball.

    This is why Arsenal is paranoid about being hacked – it’s not just the player ratings they don’t want people to see, the real issue is that they don’t want anyone seeing their data or their statistical analysis of that data.

    Bottom line: statistics is a proper mathematical discipline, with complex methods, procedures and disciplines – it’s not just a bunch of numbers (or “variables”) on which to base a subjective opinion.

    People these days seem to believe that they can challenge science with opinion. They can’t – and the only reason they’re not laughed out of the room is because, sadly, the people who listen to them don’t know any better.

    So – if someone starts quoting numbers at you, saying that it means Arsenal will get relegated or Forest wil win the league etc… I suggest you just say “Those are values of a predictor variable, what statistical analysis/tests have you done to show how it affects outcomes?”.

    I’ll post a link in a separate post for anyone with the attention span, showning the tests and when they should be applied. I doubt many got this far though – well done if you did 🙂

    1. This is a (relatively) simple guide on how to select the right statistical test:

      https://www.scribbr.com/statistics/statistical-tests/

      Each test is a piece of mathematics, the purest of all sciences, it’s not an opinion.

      But note that it is not “deterministic” (2+2 can be *determined* to be 4) it is a “probabilistic” branch of mathematics (if you flip a coin, the chance of getting 2 heads in a row can be shown to have a probability of 25% or 1 in 4, we cannot determine the result with certainty).

    2. IDontKnowWhyIcare…

      Brilliant!

      Your response is absolutely commendable!

      And just so you know, I have made it this far! *smiling*

      Unfortunately, such statistical analysis/tests as you have advocated for frequently require the use of a computer, a capability that a typical on-the-go fan lacks… Therefore, much as every user would prefer simplicity over sophistication in most products, similarly, fans would simply choose to use the predictor variable to support an opinion over the use of any other rigorous computational alternative…

      I suppose the low-hanging fruit is to advocate for fans to embrace the correct interpretation of statistics/data in relation to an opinion or argument.

      Lastly, I would say that such worthless/meaningless statistics should be curtailed at the point of generation, which is a fruitless endeavor in and of itself, sadly so.

      Brilliant argument once more!

      Cheers!

      1. Thanks Fire! And well done getting that far. 🙂

        You’re right that a computer is helpful for large data sets – and in any case it speeds things up and tends to eliminate errors.

        The penultimate paragraph was written the wrong way round btw, it should of course read:

        “So – if someone starts quoting numbers at you, saying that it means Arsenal will get relegated or Forest will win the league etc… I suggest you just say: “Those are values of an *outcome* variable, what predictor variables have you gathered data on and which statistical analysis/tests have you done to show how those variables affect outcomes?”

        Apologies, I was rushing it a bit.

  12. Statistics are like mini skirts. You see some hints but you can’t see the whole prize. Either a wonderful night or a beautiful lie.

    Btw, Mers and Kos did a lot of back passing to draw the opponents, then Almunia punts towards Adebayor, Van Persie and Walcott. It was by design by Wenger to counter defensive teams.

  13. Very well noted IDontKnowWhyIcare!

    Indeed, these are thought-provoking perspectives on a highly academic topic that many may be unfamiliar with.

    Nevertheless, I find such topics interesting, and I would normally like to participate in and sustain them…

    I hope to chat with you again, most likely via another thought-provoking article!

    Cheers! And stay blessed!

Comments are closed

Top Blog Sponsors