The most complete player of all-time?

Discussion in 'The Beautiful Game' started by y.o.n.k.o, Jan 28, 2011.

  1. Trachta10

    Trachta10 Member+

    Apr 25, 2016
    Club:
    CA Boca Juniors
    Without the existence of statistics, the world more or less knows who the best players in the world are. If you create a model and the result it gives you is random players in the top positions, you don’t say, 'Wow, I just discovered something incredible; these players are probably performing off-the-ball actions, which is why they’re ranked so high.' The conclusion you should reach is that your model is probably garbage and doesn’t align with reality. Without knowing the absolute truth, you can still get an idea of whether your model is on the right track or not.

    The algorithms of Whoscored or similar ones do align with reality because they show the players you would expect to find in the top positions as the highest-rated, so you know that, although imperfect, it aligns with reality. If you see that the best players in the world have certain characteristics in common, the most logical thing to consider is that what they have in common is what’s valuable.
     
    ganapordiego and Sexy Beast repped this.
  2. Letmepost

    Letmepost Member

    Arsenal
    South Korea
    Apr 11, 2023
    So many words wasted to arrive at the conclusion that says you basically want to keep re-arranging numbers until you get the list you have already made up inside your head in advance.

    Your past work used to be of more substance than this. It is somewhat sad. Some of the best work I've enjoyed reading into this. Less intellectual integrity with each passing moment. Sorry to be so insulting, but this is what I feel after discussing in length with you. I'm done talking.
     
  3. Isaías Silva Serafim

    Real Madrid
    Brazil
    Dec 2, 2021
    Nat'l Team:
    Brazil
    In fact, at the highest level there are rare matches in which a team does 20 shots on target. It only happens when for example Bayern faces a team from the 4th German division in the DFB Pokal. This season city had 6 shots on target per game and 3 goals per game.

    Ironically the very same argument you're using ("try until it works") also applies to shots on target
     
  4. Wiliam Felipe Gracek

    Santos FC
    France
    Feb 3, 2024
    well..I'm not a fan of Lionel Messi or C. Ronaldo... But I don't see the Why?? to minimize C. Ronaldo's great achievements. or vice versa..

    Ferenc Puskas .. Don Di Stefano .. and Edson Arantes Pelé tried to bring to the game .. the famous permanent Peak almost impossible to achieve this in long terms .. and in long careers ..but they were unable to do so .. due to the generally very precarious infrastructure and lack .. of advanced sports science. at that time ..



    But ..

    ...
    Lionel Messi and C. Ronaldo were the first players to reach the famous permanent peak.. 15 years or 20 years at their peak always.


    They changed the level of this sport forever...


    so more Respect with C. Ronaldo !




    Peak like Ronaldinho Gaucho and Ronaldo Nazario..they are seen as nothing and common... nowadays !


    But it's very much about excellence. high performance and consistency of these two football monsters

    Lionel Messi and C. Ronaldo. !
     
    Isaías Silva Serafim repped this.
  5. Wiliam Felipe Gracek

    Santos FC
    France
    Feb 3, 2024


    ..........well..I'm not a fan of Lionel Messi or C. Ronaldo... But I don't see the Why?? to minimize C. Ronaldo's great achievements. or vice versa..

    Ferenc Puskas .. Don Di Stefano .. and Edson Arantes Pelé tried to bring to the game .. the famous permanent Peak almost impossible to achieve this in long terms .. and in long careers ..but they were unable to do so .. due to the generally very precarious infrastructure and lack .. of advanced sports science. at that time ..



    But ..

    ...
    Lionel Messi and C. Ronaldo were the first players to reach the famous permanent peak.. 15 years or 20 years at their peak always.


    They changed the level of this sport forever...


    so more Respect with C. Ronaldo !




    Peak like Ronaldinho Gaucho and Ronaldo Nazario..they are seen as nothing and common... nowadays !


    But it's very much about excellence. high performance and consistency of these two football monsters

    Lionel Messi and C. Ronaldo. !
     
    SayWhatIWant repped this.
  6. Wiliam Felipe Gracek

    Santos FC
    France
    Feb 3, 2024


    ...........well..I'm not a fan of Lionel Messi or C. Ronaldo... But I don't see the Why?? to minimize C. Ronaldo's great achievements. or vice versa..

    Ferenc Puskas .. Don Di Stefano .. and Edson Arantes Pelé tried to bring to the game .. the famous permanent Peak almost impossible to achieve this in long terms .. and in long careers ..but they were unable to do so .. due to the generally very precarious infrastructure and lack .. of advanced sports science. at that time ..



    But ..

    ...
    Lionel Messi and C. Ronaldo were the first players to reach the famous permanent peak.. 15 years or 20 years at their peak always.


    They changed the level of this sport forever...


    so more Respect with C. Ronaldo !




    Peak like Ronaldinho Gaucho and Ronaldo Nazario..they are seen as nothing and common... nowadays !


    But it's very much about excellence. high performance and consistency of these two football monsters

    Lionel Messi and C. Ronaldo. !
     
  7. carlito86

    carlito86 Member+

    Jan 11, 2016
    Club:
    Real Madrid
    #1832 carlito86, Sep 13, 2024
    Last edited: Sep 13, 2024
    IMG_6793.jpeg
    IMG_6792.jpeg


    IMG_6797.jpeg
     
  8. carlito86

    carlito86 Member+

    Jan 11, 2016
    Club:
    Real Madrid
    #1833 carlito86, Sep 13, 2024
    Last edited: Sep 13, 2024
    Thierry Henry
    IMG_6798.jpeg
    https://amp.theguardian.com/footbal...-henry-arsenal-arsene-wenger-barcelona-career


    Only the armchair analyst can disregard this

    The one who saw more burgers in his life than sessions in the gym.

    The discipline of these guys will be talked for decades(and decades)

    The mind of Cristiano reminds me of this guy

    Never say die and keep pushing yourself to the absolute limit

    And when you reach that limit you raise it even higher.


     
  9. Wiliam Felipe Gracek

    Santos FC
    France
    Feb 3, 2024
  10. Wiliam Felipe Gracek

    Santos FC
    France
    Feb 3, 2024


    ....
     
  11. Wiliam Felipe Gracek

    Santos FC
    France
    Feb 3, 2024
  12. carlito86

    carlito86 Member+

    Jan 11, 2016
    Club:
    Real Madrid
    #1837 carlito86, Sep 14, 2024
    Last edited: Sep 14, 2024
    IMG_6806.jpeg

    IMG_6805.jpeg

    IMG_6802.jpeg

    IMG_6803.jpeg

    IMG_6801.jpeg
     
  13. Sexy Beast

    Sexy Beast Member+

    Dinamo Zagreb
    Croatia
    Aug 11, 2016
    Zagreb
    Club:
    --other--
    Nat'l Team:
    Croatia
    #1838 Sexy Beast, Sep 14, 2024
    Last edited: Sep 14, 2024
    That actually says quite the opposite of what you think it says..

    If there are 100 shots on target, 60% xGoT means that player eill score 60 goals.. so if we include goals in the table, by doing so we are already including 60% of shots on target. (40% shots on target are excluded)

    On the other hand, if 30% of key passes end up as assists, by including only assists in the table, we take into consideration only 30% of key passes. (70% of key passes are excluded)

    So the point is what we are leaving aside if we include only goals and assists.

    We are leaving out 40% of shots on target that dont end up as goals and have questionable value and impact for the reasons stated above that you've agreed with. And we leave 70% of key passes that dont end up as assists that have exclusively a positive value.

    By including only goals we destill the most value that positive shots on target can provide. By including only assists, we are overlooking a tremendous creative effort that doesnt end up as assist.

    40% of shots on target that are not included have questionable value because we dont know what kind of shots those were, while 70% of key passes not included have an accumulative, positive value, because key pass can not be a negative thing.

    By only looking at goals and assists you are massively downplaying creative efforts.

    But by including goals, assists, shots on target and key passes, you are introducing a random element in a form of shots on target and potentially rewarding player for missing big chances.

    Count goals and key passes and you get the best of both worlds.
     
  14. Sexy Beast

    Sexy Beast Member+

    Dinamo Zagreb
    Croatia
    Aug 11, 2016
    Zagreb
    Club:
    --other--
    Nat'l Team:
    Croatia
    +/- model is an absolute dead end in football. Yes. There are 21- players on the pitch at any given moment with variable performances that much more influence what is going to happen on the pitch than that one player in focus.

    The model is capturing "random" patterns in matches that have nothing to do with performance of that one player.

    It is completely useless model. Even worse, it is misleading model because it will make you think you are capturing some pattern while you are not.

    Football is way too complex for such simplistic models. Even basketball is pushing the edge of it, which is why no one seriously takes +/- model in basketball.. and basketball only has 10 players on the court.

    Capturing value of off the ball movement is an extremely difficult task and doing so would revolutionize statistics in football. What you are suggesting is far from it...

    Still by far the best way is to measure on the ball actions and keep in mind that it doesnt capture off the ball actions in its entirety.

    Nowhere in the on-ball actions analysis it says that this should be the football Bible. Outsourcing your opinion on what "stats say" is wrong no matter the methodology.
     
    Trachta10 repped this.
  15. Sexy Beast

    Sexy Beast Member+

    Dinamo Zagreb
    Croatia
    Aug 11, 2016
    Zagreb
    Club:
    --other--
    Nat'l Team:
    Croatia
    Dont get discouraged by naysayers. You are absolutely spot on.
     
    Trachta10 repped this.
  16. Sexy Beast

    Sexy Beast Member+

    Dinamo Zagreb
    Croatia
    Aug 11, 2016
    Zagreb
    Club:
    --other--
    Nat'l Team:
    Croatia
    #1841 Sexy Beast, Sep 14, 2024
    Last edited: Sep 14, 2024
    The thing with off the ball movement is that is team- and tactically-dependent. There is no such thing as 100% good off the ball movement from a point A to a point B. It completely depends on the context on the pitch - positioning and movement of teammates and opponents.

    Off the ball movement is a lot about sticking to assigned roles and tactical patterns rather than being a product of improvisation and ingenius idea.

    There are also many complimentary styles of movement and these almost always work in conjuction with each other.

    If a player A makes a certain movement off the ball, by itself is not much and has no inherent value. It might only work if his teammates B makes a certain complimentary movement.

    Obvious example would be if two attackers make a run towards the far post for a cross. In this instance, they are not working well together and likely won't score because they are bring two defenders into the area with their separate runs..

    The run to far post works only if the other player makes a run to near post for example..

    Also the most value off the ball movement brings is captured with stats like goals, which is what few of the members are fishing for for Cristiano in here.

    The "uncaptured" off the ball movement in question are actually rare instances worth exploring but are not game-changers.

    In the case of Muller, his off the ball movement is already captured to msot degree with his goals and especially assists. He has a high rate of assists precisely because of his off the ball work and intelligence. So it is captured.

    We are really talking about off ball actions that are niche and that could swing things left or right statistically, but wouldnt result in a complete rewrite of stats and footvall algorithms.
     
  17. Letmepost

    Letmepost Member

    Arsenal
    South Korea
    Apr 11, 2023
    If you want to engage in a conversation about a comprehensive debate about why the Plus-Minus model as a premise is flawed from the start. Let's have it, I'm itching for an in-depth conversation about it as a novice.

    So far, I've had three discussion points.

    1) The model failing to account for the differences across leagues and era.

    -> I asked what is so intrinsically mathematically flawed about what the model says about the players from the same squad, same league, and same era, across multiple years with 10,000 minutes of sample size or greater.

    I got zero response, and got a weird circular ramble about the final list needing to satisfy the maximum number of people who appreciate famous players, and how WhoScored should be the standard since it does that so well. All the biggest fanbases are covered.

    2) The model having high variation, and requiring extra-large sample size.

    -> I said that's more likely an issue with difficulty of proper execution, not doomed scenario from the get-go. Unless the variation is impossible to account for with changes in the formula and extra-large sample size.

    We are not discussing season-long sample sizes with roughly 3,000 minutes of playtime to draw from. These are 10,000 minutes sample sizes or greater. If this is insufficient, I am willing to go all the way to full careers.

    Is that doomed also? Is this speaking from experience, or after many failed attempts? Abolsutely zero chance possible solutions at all? Resolutions coming from a casual fan? Of course not, it was never what I asserted. I said the premise was fine, and that the execution of it was lacking.

    The discourse over a minimum threshold of sample size, and accounting for as many confounding variables as possible never took place. I had more conversations about Thomas Muller, than any insight into the formula and what makes it doomed to begin with.

    3) Off-the-ball actions being accounted for by the Plus-Minus model better than WhoScored algorithm.

    -> I said there was a likely possibility. Trachta10 said such an explanation was not only wrong, but that we should go back to re-hashing WhoScored algorithms until the maximum number of the fan-base gets a necessary number of famous players with numerical values that satisfies their general curiosities.

    If you are using this conversation as a proxy for whatever battle of wills you have versus Cristiano Ronaldo fans, count me the ******** out. I do not care what algorithms gets the maximum reach from the Lionel Messi, whilst managing to trigger just the right number of Cristiano Ronaldo fans for just the right number of exposure and reach. Let Trachta10 perfect that art.
     
  18. Letmepost

    Letmepost Member

    Arsenal
    South Korea
    Apr 11, 2023
    As an additinal request, please leave behind whatever worries you have about your favourite players not being well-represented in the process of working out the formulas. I truly cannot take any more of these personal worries about anything that might lead to a slight of any nature versus your personal heros. I can assure you, that your personal heroes are indeed gods that stand above these flawed mathematical equations. Now let's continue the conversation.

    If that is impossible, just leave them out of the argument entirely, and work with the numbers of players from teams you have zero emotional attachments to. Please, anything but circular arguments about the conclusions not being appetizing for a fan of Lionel Messi, Diego Maradona, or whoever the ******** it is that you worship.
     
  19. Isaías Silva Serafim

    Real Madrid
    Brazil
    Dec 2, 2021
    Nat'l Team:
    Brazil
    Again you're comparing key passes with assists when my argument is comparing key passes with xA and shots on target with xGoT. As I demonstrated, the best playmaker in the world has 0,14 expected assists (not actual assists) per key pass so by counting all his key passes we're rewarding him for passes that on 86% of cases aren't likely to turn into goals (far from the goal or with a lot of defenders between the finisher and the goal or from angles hard to finish or a combination of those factors. hence, very low quality key passes).
     
  20. Isaías Silva Serafim

    Real Madrid
    Brazil
    Dec 2, 2021
    Nat'l Team:
    Brazil
    The exact same thing applies to movements with the ball mate. If player A makes a through pass he will depend of player B to make an off the ball movement and catch this pass. If player A makes a cross to the box he will depend on player B to finish them by out jumping and out running the defenders. If player A tries a dribble he will depend on the defender miss the tackle, etc...

    Now let me ask you something: Why is a chance created via “through balls/dribbled players” seen as more valuable than chances created via running behind or receiving the ball in space virtually bypassing the defense?
     
    SayWhatIWant repped this.
  21. SayWhatIWant

    SayWhatIWant Member+

    Jan 10, 2015
    First, what you are describing is chances - scoring chances - not actions - so you did not follow up your message addressing the primary complaint. Again, I challenged your unsubstantiated (and incorrect) assertion that more actions = win.
    Football is a complex, nigh impossible game to model and predict. There is a great deal of game theory involved. The idea that simply doing more of an extremely heterogenous set of quasi-arbitrarily defined actions is a marker of win probability is outrageous.
     
    Isaías Silva Serafim repped this.
  22. Trachta10

    Trachta10 Member+

    Apr 25, 2016
    Club:
    CA Boca Juniors
    #1847 Trachta10, Sep 15, 2024
    Last edited: Sep 15, 2024
    Let’s see, where do you get the idea that someone 'adjusts' the numbers to make their favorite player rank higher? Where do you get the idea that whoscored or similar platforms are dishonestly creating models with the goal of, let me guess, Messi always being the highest-rated? Because that’s what you’re doing, you're talking about a conspiracy theory.

    You’re the one who doesn’t like the result, and that’s why you’re thinking about alternatives.
    Ironically, you’re accusing others of fixing the numbers until they get the results they like.

    I suppose you understand what I mean when I say, 'the list your model produces has to align with reality.' It’s not about 'wanting my favorite player to be at the top,' it’s about wanting my results to make at least some sense and coherence.
    Because if a model that you consider good has Thomas Muller as the best and players like Cristiano Ronaldo not even in the top 20, obviously the first thought that comes to my mind is that the model is trash. Coincidentally, when you start to think about it, you see that this Plus-Minus model has many flaws. That’s why I had the boldness to criticize it and point out the areas where it didn’t make any sense to me.


    The point here is, how do you figure out what the absolute truth is, more or less? I think the most logical approach would be to analyze the best players in the world (obviously we're talking about something subjective) and from there start identifying common patterns they share, and then translate that mathematically, with the limitations that implies, as the complexity can be very limited.

    Something I believe the best players have in common is that they perform many 'positive actions.' For example, Messi is probably one of the players who generates the most shots, key passes, dribbles, passes, etc. This is no coincidence, which is why it makes sense to value these actions.
    It’s quite evident that the best players are those who have the most influence on their teams. A common pattern I find among the greatest players in history, like Pelé, Maradona, Cruyff, Puskás, Di Stéfano, and also Cristiano, is that they tend to be very 'gravitational' within their team. How? By assuming more responsibility than the rest. How? By ensuring the ball goes through their feet more often. And how does this translate statistically? (with the limitations that entails) Well, these are the players who generate the most shots, key passes, dribbles, and passes. So yes, one aspect that makes you good as a player is performing many positive actions with the ball; that’s the logical conclusion.

    Obviously, I know this is much more complex, but in general, basically speaking, it’s a pattern that the best players in the world share.
     
  23. Letmepost

    Letmepost Member

    Arsenal
    South Korea
    Apr 11, 2023
    #1848 Letmepost, Sep 15, 2024
    Last edited: Sep 15, 2024
    I truly don't give a shit who comes out on top. Lionel Messi can top the perfect version of the Plus-Minus model for all eternity for all I care. If the hypothesis of the premise is sound (which this might be), sample size is sufficient, and all the confounding variables are accounted for, and the most apt analysis model is chosen for the data, and then many versions are carried out for the same premise, to see what possible errors could have been previously forgotten about. Why not? How many of these topics have you mentally computed inside your head? You probably spent more hours computing just the right ratio of statistics for maximum coverage of the fan-bases.

    I am not going to waste time. Either discuss the model in an intellectually honest manner that addresses my points, or stick to making "mathematical" models that consist of you going over OPTA data, and using appeal to authority. It is people like you that actually cheapens Lionel Messi's legacy. You have become a snake-oil salesman when he clearly does not require such help, other than to win arguments on the net. Nothing you do is helping me find out anything insightful at the moment.

    Why not just assign values to mimic Ballon d'Or to 100% accuracy at this point in time? Wait did their street-cred fall compared to WhoScored. Check what's trending on Twitter for maximum coverage.

    Go back to doing what you are best at, and stop annoying me. Thank you for your input, and I would have remained civil and greatful for what insight you gave, if you didn't argue in such a non productive manner. At this point in time, I rather spend time thinking about this topic alone than arguing with you, and getting emotionally vexed.
     
    Gregoire and Isaías Silva Serafim repped this.
  24. Trachta10

    Trachta10 Member+

    Apr 25, 2016
    Club:
    CA Boca Juniors
    You win by scoring more goals than the opponent. I’ll only talk about attack—how do you score goals? By taking shots on goal. Without a doubt, your goal should be to generate as many shots as possible on the opponent's goal. I think we all agree up to that point
    In general, I would prefer to have players whom I’m sure know how to create scoring opportunities and can do it many times. I’m simplifying, but for example, a player who dribbles a lot doesn’t do it just for the sake of it; rather, this action allows them to, for instance, make a better pass because they are now in a better position, or to be in a better position to attempt a shot. So a player who dribbles a lot increases the team’s chances of taking a shot on goal. The same can be said for playmakers. A player who is a great passer, who can read the moment of the play, and who can do this many times, is valuable because they also increase the chances of making more shots and thus scoring more goals.
    When I use the term 'positive actions,' I’m referring to this. Obviously, I’m focusing on the individual aspect; there can also be collective and tactical actions that I couldn’t attribute to a specific player but rather to the team as a whole.
     
  25. Trachta10

    Trachta10 Member+

    Apr 25, 2016
    Club:
    CA Boca Juniors
    Suddenly you got very sensitive, relax, bro. Honestly, I don’t understand what you’re talking about. Think carefully—I didn’t say the Plus-Minus model was bad because Messi and Cristiano weren’t at the top of the list. I mentioned several points and provided I think valid arguments for why the model had many aspects that introduced randomness or were intrinsically unfair, for example, assuming all opponents are the same, favoring teams that score more goals, etc. In other words, I made an honest critique.

    It's not that I’m saying, 'if Messi isn’t at the top, then the model is useless,' although I do believe there is a certain 'per se' idea of what one should expect to find.
     

Share This Page