Pandora’s Box: Statistics in Football

Nate Silver: Rock Star
Nate Silver: Rock Star

I like Nate Silver. I read his political blog virtually every day and followed it religiously during the 2008 presidential election cycle here in the US. He’s a man who knows his statistics and also his sports: he writes for Baseball Prospectus and uses more decimals than I do. A lot more. So yeah, I like Nate Silver and when I learned that he was involved in ESPN’s Soccer Power Index (SPI), I was pretty excited. After all, Nate has a bigger brain than I do and maybe I would be able to learn a few things about approaching football/soccer statistically.

Nate’s article is here. The actual SPI rankings are here and you can the full methodology here. You can read about how smart the guy is here.I’m not here to bash the SPI or anything like that. I think it’s probably being bashed enough around the footballing world as just another American attempt to force the beautiful game into a box that is easily understood via a spreadsheet and anyway, Nate himself writes that the “SPI is designed to serve as a general guideline” rather than a one-stop-shop for who is going to win the next World Cup.

Basically, I can’t quibble with the numbers involved or how they’re done because I don’t have the background to truly look at them, but I do think that there are some parts of this that are worth discussing, even from a layman’s point-of-view. If I understand it correctly, the idea behind SPI is to rank who is most likely to win the next World Cup or at least if you put two teams together determine who is more likely to win that match (which might sound contradictory to my previous statement that it’s not designed to do that, but it’s in the semantics: most likely means it’s about predictive power rather than “Team A is the best in the world”…I hope that makes sense).

This is what the methodology says:

The SPI proceeds in four major steps:

Calculate competitiveness coefficients for all games in database

Derive match-based ratings for all international and club teams

Derive player-based ratings for all games in which detailed data is available

Combine team and player data into a composite rating based on current rosters; use to predict future results.

I added the emphasis. There are, of course, lots of numbers (and decimal!) involved in this, but I highlighted what I think might be the key problem. Nate writes, “…basically, we’ve taken results from every recent game in the four key European leagues (England, Germany, Italy and Spain), plus the Champions League, and assigned credit or blame to the individual players on the pitch based on the results of those matches.” This “credit or blame” is what’s bothering me.

If you’ve read this site for long enough (or its original iteration at The Offside), you’ll know that I worked for a little while on just such a system–though certainly with less brain power and fewer resources–so you’ll also know that I’m pretty open to these ideas. It’s just that after working on these things for a while, I realized, as some commenters had pointed out early on, that there really isn’t a good way to measure football statistically for each player. You can take teams and figure out a lot of what’s going on in the games through the total numbers, but assigning individual blame or credit is a bit misleading for several reasons. The first is that there are questions about the stats themselves. Someone must be subjectively judging whether or not a pass was on target or not, whether the player receiving a pass was in the right position to receive the pass or not, etc. There are so many variables that I find it basically impossible to adjudge how a simple Player A to Player B pass should be construed statistically that once you add in 9 other players on that team and 90 other minutes of action, it’s basically impossible to figure out who did what and was better by the numbers. That’s why watching the match and getting an understanding of the system and who did well within those particular 90 minutes is so important (and part of what makes Kevin’s reviews so important and difficult).

The SPI gets around this whole thing rather handily: “The first and foremost requirement is that the sum of the ratings for all individual players on a team must equal their team’s rating for that game; soccer is too much of a team sport to make any other assumption.” Fair enough, I suppose, but I still think it presupposes that the stats are being kept in regulated, methodical, and standardized ways. I’m not so sure they are. For instance, how do they know how far Xavi ran on a given day? How do they track that, exactly? And is it different in La Liga and the Champions League or in internationals? Are they like NBA stats keepers?

Given all that, what’s the point anyway? I’m not a big fan of “objective” polls (I hate the BCS with a burning passion) because I don’t think they’re any more objective than the subjective polls that at least admit to not being objective. I care about NCAA men’s basketball rankings only insofar as they get my Kansas Jayhawks (ranked #1 incidentally) into the tournament with a better seed, but I don’t like it and I can’t really get behind any Ratings Power Index (RPI) for any sport (NBA springs to mind). RPI basically tells me what I already knew as a fan of the sport: what teams are good.

I know what teams are good in international soccer. If I were to blindly rank teams based purely on the speculative nonsense that runs around my head after a long weekend of watching internationals, I’d come up with a list that is basically the same as the SPI. It wouldn’t be exactly the same (I think Spain is better than Brazil right now, for instance), but it would certainly be within the realm of the understandable. After all, compare the FIFA rankings with that of the SPI. They are very similar (they have 8 of the same top 10) and yet also different (Italy is 4th according to FIFA and 13th according to SPI*)–in fact, the SPI looks a lot more similar to the ELO rankings, but they’re all basically the same lists in slightly different orders.

So, I appreciate what the SPI is doing because I like the idea of being able to understand more about the game through statistics, but I’m not sure what it really does. I’d be okay if FIFA adopted this system instead of what they have for the seeding of teams in the World Cup, but I don’t suppose it really matters either way since both systems are going to be flawed in some manner and I don’t have the brainpower to compare the two outright.

Take what you will from the SPI, but I’ll be checking in on it as well as on the Castrol Rankings that Kevin discussed in a previous post. And yeah, I’ll hopefully be writing up something about those rankings later this week, but suffice it to say that I’m not a big fan already. Long story short: the SPI has lot of decimals. I like that.

*One reason for that might be SPI’s updates compared to FIFA’s. November 20 will be the next time FIFA updates their rankings (after the WCQs are all wrapped up), so it might be better to compare the two then rather than now.

By Isaiah

Isaiah is a co-founder and lead writer for Barcelona Football Blog. He currently lives in the greater Philadelphia area.


  1. You almost Hectored my longest comment even in the previous post! 😀

    I will Hector you, my way in return, by not reading your post till I go and have my dinner first!

    1. in the case you dont see my reply to your comment in previous post:

      i dont think pep really like the idea of having yaya as a box to box midfielder,just take a look at the games that both yaya and busi play together,yaya always plays DM.and i dont think thats because pep just dont trust busi as a DM,couse even against weak teams like cultural or whatever he played yaya there.
      but ofcourse you may think its because of busi’s crappy performances that pep always play yaya as a DM not yaya’s inability to play as good as keita as a box to box midfielder.but even last year which busi was playing pretty well he didnt use yaya as a CM,but as a CB.i think pep trust yaya’s defensive quality far more than his ofensive qualities.
      yes he have played CM in monaco but comon,monaco is NOT barca.there must be a reason that after a year and half,pep only used yaya as CM once(against man city in pre season)

    2. Busi’s crappy performances!? If Pep thought in those lines Busi wouldn’t have played more than Yaya in the league don’t you think!?

    3. Actually, Pep playing Busquets because he is not performing well. What?!

      As I mentioned in “That comment” Pep is trying to win busquets back to form before January, just in case. There is only one way: playing him more.

    4. Ramzi agreed with what you said on the other post as in how we won’t have to sacrifice on quality while going for the rotations with having Masch in the squad… But I think that it’d be better to find a replacement of Keita, a BtB with probably more defensive quality a.k.a Yaya Toure, instead of going for a pure DM.

  2. *
    Have a look at Keirrison’s second goal for Benfica… Both of them he scored only in friendlies, but this one is just an AWESOME finish 🙂

    Sport says we are being linked with Philipp Lahm again, but didn’t Pep already say he’s been following Cristian Ansaldi for some time now!?

  3. Nate may be smart but if he participated in ranking Argentina(with the maradona factor) in 5th, Uruguay(has not qualified) in 8th while leaving Paraguay in 20th, than he can’t be THAT smart. I guess I will look at the SPI rankings like I look at 99.9% of ESPN Footy articles, like they were written at, in other words, like a joke ie. England 3rd hahahahahahahaha.

    I say Isaiah should have a blog at ESPN. All in favor say “I”.

    1. but then he would be posting less here, and it would be harder to get to the articles and there wouldn’t be sweet barca pics surrounding the article. Plus, he would have to be less biased. I oppose this idea. And i was thinking the same thing about teams that didn’t qualify

    2. You have a point but than Isaiah could quit his day job and just do ESPN ( given he got paid a lot of dough) and this blog and therefore have more time to post articles here not to mention he would have more resources and ESPN would be more tolareable.

  4. Well Isaiah…Regarding the topic, We discussed this once before. May be its me resisting change, but every time I read about a statistical attempt like this I only imagine people understanding the game like this:


    And analyzing it like this:


    “there really isn’t a good way to measure football statistically for each player. You can take teams and figure out a lot of what’s going on in the games through the total numbers, but assigning individual blame or credit is a bit misleading for several reasons.”

    That’s exactly where I do not agree. I am not a statistics guy. But from a football point of view its even easier to analyze an individual performance (or even predict how he will perform based on perimeters) through statistics than to analyze a team. I do not want to be repetitive in this matter.

    For the teams performance/predicted results, it depend a lot on the kind of teams they will play against. Something that will only be defined by a draw (no science in that). For example, if Spain meet Netherland, Brazil, and Portugal in the world cup knock out stages, they may have a better chance to win than meeting England, Ivory coast, and Germany. Saying so, it doesn’t mean that the first three are better weaker than the following ones, but its because the teams character of England, Ivory coast, and Germany works as Antibiotics for a team like Spain. Meeting one of them is something. Meeting them all is something else. While Netherland, Brazil, and Portugal are great squads but their style and character are conformable to that’s of Spain, so quality will decide.

    All that and we are only giving examples of big teams. Imagine how it may differ if Spain meet a weak team in one of the knock out stages, or if they meet Argentina. Not only on their chances to pass the stage but on their ability to pass without being extra exhausted, without injuries, or booking. All that will have an impact on their result on the following stage, then the following to the final.

    1. So, first off: do you like or dislike the SPI?

      What I think you misunderstood about the sentence of mine that you quoted was that I wasn’t talking about analyzing the performance, but rather the actual statistics.

      For instance, if a pass is made and it’s off target, that’s one off-target pass for the team, regardless of whether or not it was the passer or the receiver’s fault. It’s much easier to define the stats that way (team stats) than to define, statistically, whether Player A or Player B was at fault (individual stats). That was my point, not the question of analyzing the actual contributions across the board of a team versus a player.

      Even so, I’m not sure I agree that it’s easier to say individuals performed better within a particular game or a team. Take Bojan’s performance against Cultu in the last CDR match. Kevin didn’t think he worked well within the system while I thought he worked very well, but we both agree that the match was a good team performance.

      Obviously, the question is stylistic in the end (Nate Silver makes the point that Italy is less likely to be upset by weaker teams because of their inherent lack of goals allowed and that they win basically by attrition while a team like Spain is susceptible to upsets because they play a much more open style)–but I don’t know how you can define styles in numbers very easily.

      Naturally I hope that FCB gets to the top of all these rankings (in the SPI methodology they’re equated to Brazil in the Country:Club analogy), but it doesn’t really mean anything except on the field as we kick some ass.

    2. the more i think of yaya playing keita’s role the more i dream to see it!
      i love keita, but i think THE YAYA would be atworst twice as better in that role then keita. hes better defensivly and offensivly.
      with keita being 30, yaya would take his spot. leaving a space for a world class argentine defensive mid.

  5. …”So, I appreciate what the SPI is doing because I like the idea of being able to understand more about the game through statistics, but I’m not sure what it really does.”…

    I can’t say anything better 😉

    wow those nate silver links were interesting kevin!!

    football is one of the most chaotic sports ever and the system would need to have like thousands of stats (variables in spanish) to measure it properly… and surprises would happen still

  6. What do you think of NFL power rankings, Isaiah? Agree that Colts should be number one right now?

    1. I could care less about NFL power rankings, to be honest. I would imagine Colts would be first, yeah, but the Saints have a pretty decent shout for it too.

      Fortunately for all of us, there are the playoffs to give us that “ultimate victor”, though I would hotly dispute that it actually does that. It would be pretty cool if there were a way to get all the NFL teams to play each other home and away, just like they do in, you know, La Liga…THAT is the ultimate RPI.

      That said, I also love stats in sports and love the concept of making these things better. Obviously I love stats. We have a whole category for them here. I’ll definitely be trying to keep up with this.

  7. First off, I do not dislike SPI, it’s kind of cool to read, but it’s useless (May be a harsh word but let’s put it the way it is). And I will be harsh for purpose as I believe putting stat in the right place will be a revolutionary step in football development. Instead there is a kind of determination (with or without intention) to abuse statistics to serve the betting game. That’s the whole thing isn’t it? This team will most likely win, so it’s there where you need to put your bets. Really I can’t see anything more about it.
    It may be a nice tool for fantasy leagues, Football manager game, or so. For real football it doesn’t deliver anything special, to say the list. If gathering a council of the most brilliant brains with supercomputer computers and decades of effort creating a methodology to tell me that Spain is more reliable to counters than Italy, then, hm…They could have asked me from the beginning. I would have saved their time, money, and brain cells consumption. Here the way it goes: Italy plays Zonal defense defending in their own half, when they lose the ball they don’t seek it anymore. Instead they run back to their own half repositioning their troops and closing gaps. Spain instead do not do that. Small teams usually play counter attacks, they need space. Against Spain it exists, against Italy it doesn’t. Piece of cake. Notice, I didn’t use any number, and I came up with an explanation in the very same day that you posted your article. Well, they can earn the credit of reinventing the wheel though. But so they know, Netherlands is more likely to be beaten against a team that plays 5 midfielders than England. Just as an example to save some more decades of work. Stat is a new thing in football. To prove itself it has to tell me what I don’t know already. If while Pep is busy in training some genius stat guy will come with a two tons folder that prove Valdes being a better keeper to start for Barcelona than Pinto, well…I can predict Pep reaction. No reaction, as he is too busy for that.

    But based on the same SPI standards (without customizing it the way holly book meanings get customized everyday to meet the new scientific facts), I challenge them to explain why Denmark won the Euro 1986 when they were gathered from their vacations to be involved after kicking Yugoslavia out. Greece? Italy was beaten by Korea and knocked out of the world cup. Recently Egypt was out of the competition to qualify to the following world cup, even Egyptian fans knew it. Out of sudden all the other teams started to lose points like crazy and last night there was a game between Egypt and Algeria. Egypt needed to win 2-0 to dictate a third deciding game. Or else Algeria qualify. Egypt won 2-0 and tomorrow they will play in Sudan. Analyze the group process from beginning to end and calculate that.
    SPI system claims that it can predict future performance and the easiest way to test that is to take a slice of time from the past, apply the formulas on the first extreme of the time segment, and compare the “Calculations” output on what actually happened on the other end. Would the SPI formulas lead to predict Italy to win the last world cup if we used the formulas using the data existed before the world cup? Would Italy win the world cup 1982 instead of Brazil? Greece winning euro? Argentina almost getting knocked out of the following world cup? Sorry to be harsh again, but its just a waste of time.
    It’s a shame, because statistics turned to be more descriptive than constructive when it comes to football. The SPI ranking for the teams does not differ than the FIFA traditional ranking that they were issuing since decades. They used to base their ranking on teams results, as simple as that. The team win more, get more points on the rank and head the table. What the SPI doing is the same. I can see them creating formulas, the result turn to be that “Cyprus will win the world cup” they laugh out loudly! “Impossible! But you know what, if we change this variable and calculate it this other way, it wil make Spain and Brazil come first, which is more realistic. So yea that’s better”. Its more customizing the equations to express the facts than to analyze it.

    Put chips in players boots (Happening), and any kind of sensors you need. Let the players run, jump, reflex ect…and measure their initial attributes. Then if you have a player (A) passing the ball to player (B) in training:
    Based on players positions and Player (B) initial pace and initial speed when the pass was executed, define where it is the perfect point for the receiver to meet the ball, which lead to tell who strong player (A) has to pass the ball so it reach that point in the right time. When the process end, analyze it and put it in the database. The gathering of such data can give massive info for a coach (specially in youth academies):
    – If the ball was played too fast, then the player who is passing the ball either has a problem reading his teammates qualities (which is essential to improve) so he played it too fast for the receiver. Or the player nervous system is not sending the accurate message to muscles from the brain (the passer know how fast he has to pass the ball but when he pass it end up being too strong). The coach will know what to improve here. There are specific drills for each case.
    – If the passer plays the ball accurate in front of a player that is yet to run but fails to pass for a player while in the move, then that’s a different issue to fix.
    – If the play who passes played the ball with the right speed but wrong spot, then we need to improve that detail.
    – If the pass is perfect but the receiver didn’t predict it in the right place then we know what to improve.
    – If he reached the right place but couldn’t control it, then again we know what to improve.
    I can keep going forever with this. Actually every football technique could be break down to subs. For example, analyzing the player execution for every technique tells where you need to improve in order to perfect. All that kind of data will not be valid through one event. It need to be built as a data base where Player (A) pass the ball 1000 time during training from different situations, then all the data get analyzed and passed to the coach as a meaningful report that will raise his eyebrows. It goes the same with games as well. And that’s where investing stats to analyze individuals is more realistic than exhausting it by analyzing teams.

    1. Ramzi, we live in a world where some statisticians are soccer fans. And, when you have a statistician, who is madly in love with soccer you get SPI. Love it or hate it, people will quote and stand by it more than what you or I say, simply because it is backed by numbers and not by brain cells. Peace!!!

  8. Correction: By mistake I typed that Denmark won the Euro 1986 (I dont know where did i get this number from) it was actually 1992.

    And I noticed some funny spellings in the comment as well 😀

  9. You can’t quantify a game with numbers any more than you can count snowflakes.

    Does the college football poll accurately capture which team was best in the U.S.? No. Could a playoff system? Sure, but only with enough games to overcome statistical anomaly, a.k.a. an upset.

    Hockey, baseball and baketball, in the U.S., rely on a 7-game chamionship series. The best team usually wins. Not always, but usually. With one game, shit goes crazy.

    In attempting to use numbers to quantify performance, as the Castrol and SPI are doing, it’s going to be messy. If Xavi misses a pass because Keita cut the wrong way, whose fault is that?

    What statistics do is identify tendencies, based on a number of variables. They aren’t always right, but more often than not, tendencies identify potential. In that, there can be value.

    Does this exclude oil/water combos? Nope. Look at us v Rubin Kazan for an example.

    1. “You can’t quantify a game with numbers any more than you can count snowflakes.”

      To an extent you can. You can quantify a game as completely as you can determine statistical catagories for which you want to keep records. Theoretically the more variables you have the more complete your model is and thus more accurate, but no matter what in the end it’s a statistical model.

      “Does the college football poll accurately capture which team was best in the U.S.? No.”
      Does the college football poll use statistical models? I’m not saying such a system would correctly predict the winner, but such a statement seems irrelevant.

      “Could a playoff system? Sure, but only with enough games to overcome statistical anomaly, a.k.a. an upset. Hockey, baseball and baketball, in the U.S., rely on a 7-game chamionship series. The best team usually wins. Not always, but usually. With one game, shit goes crazy.”
      Of course, this is what a statistical model does. Could you flip a coin 50 times and have it result in heads every single time? Yes, of course. The probability of this is 1/2^50. With one game you will have the exact same probability to win or lose no matter what. If you flip a coin 49 times and it’s been heads every single time, what’s the probability of flipping a coin to land on heads? 1/2. Trials are independent.

      I’m not entirely sure we’re in disagreement, but I think that we have a different view of the idea of what a statistical model for football could do.

  10. What’s truly funny about the whole situation is that when sabermetrics first became an idea the baseball community took some time to warm up to it; a fact that is still somewhat true to this day!

    Game theory and statistics have the possibility to provide predictive analysis to most any system. The question is not whether or not it can be done, but maybe more whether or not it can be done yet. The key is in the math, and this might be particularly true in football. In baseball you see the result of many individual trials. By inning, by at bat, even by each individual pitch. You see the formation of set “plays”. Football is much more chaotic. This DOES NOT make it impossible. It just makes it harder.

    Do I believe that this is as reliable of a statistical system as those that Nate Silver has previously pioneered? No, I don’t think it is. However, I don’t think that this is his fault or something he can fix right away. Football has to be the absolute worst game system I have ever seen for record keeping. It’s very very subjective. I think it’s going to take a lot more dedicated record keeping and formation of more statistical catagories before the system becomes complete, but that was true for baseball as well. I think in general America is much more in love with quantizing individual performance while the rest of the world wants to romanticize the idea of the player as a some sort of larger than life idea.

    The fact that he was able to create a system so quickly that was able to come reasonably close to established rankings is an accomplishment itself. I hope he continues work on the system and that it’s somewhat premature to write the whole thing off as doomed to failure.

  11. “America is much more in love with quantizing individual performance while the rest of the world wants to romanticize the idea of the player as a some sort of larger than life idea.”

    Well let’s agree then that Statistics can contain football (Players) when it can predict accurately the Death dates of people (Life). Some people feel that numbers tell all the stories. Probably including poetry and love as well.
    I think Americans (without generalizing) failed to understand this newly introduced game. I remember during the world cup in USA when the football pioneers there promised that USA will win the world cup 2010 (I can imagine how I will have to eat my words if it happens, but then it will be against their own Stat horoscope). I don’t see lot of difference in the way football is approached there at the moment. No offense intended.
    As a footballer who has some knowledge in this game, I can say it BOLDLY : The day when the teams will only play in the domestic leagues and win the CL without playing it (if we can calculate the winner in advance, why rewinding the process by playing each and every game?) will never come. Nor will it happen that National teams will play only qualifications before calculating the world cup winner. I know no one is targeting that level of numeritizing the tackles and subtracting the shots, but the mentioned objective -if honestly targeted- must have this unlimited limit at least in the statisticians’ fantasies that they don’t dare to share.
    Yes these numbers may meet the actual results sometimes, but so as the predictions some duded do at the beginning of every year regarding what is going to happen in the following 12 months (You know…A president will die, a volcano will explode, and a fish will jump out of water to swallow an elephant). At the end of the year, the believers only remember what actually happens “Get out of here! How can’t you believe in this! He said it 3 months ago that there it will rain this winter!” , mind you, these dudes also use numbers and calculations. When the stat says that Spain and Brazil are the strongest at the moment for example. No shit! 😀 But again, what kind of analysis would have told that Greece and Denmark are going to win euro cup? If you take all the big competitions that’s based on knock out rules, the surprise was only not ending up with a surprising winner. Can we say the same about other games (especially in the US)? I don’t know, and I don’t follow. If these games are as predictable to be contained in numbers then I am not even interested to follow.
    If Brazil plays against Spain 10 times on daily bases, you will never be able to predict who will win the 11th game. Period.
    If I put water in a bottle and start shaking it, it will never turn to juice. No matter how long will I keep sayin:”Oh wait, lets try to shake the bottle faster”…”Oh, May be I need to jump on one leg while shaking it?”. Water is water, and juice is…not water. It’s not that the stat analysis output is not accurate because we are still in an early stage; it’s not true because the approach is wrong. “Wrong way”, “Dead end”, “Just forget it!” 😀

    Talking about the objective of this whole mess, what’s the objective to put all this effort? Really. Predicting who will win against whom? Well let’s say-just saying- that we will reach that point. So what? I mean, seriously…
    It’s more a cultural habit than anything else. One of the rare cases where numbers oppose common sense and science oppose pragmatism.

  12. Plus statistics don’t take into account the true variable, a single playe having the match of his life, or a defender cocking up a clearance and his side scoring a goal that completely changes a match.

    It’s why I say that on paper, statistics can present a trend, or possibility thereof. On paper, Brazil was better than France in the ’06 World Cup. And yet, they advanced.

    Statistics can say that based on x or y variables, this team will beat that team. Drop the ball, and statistics go out the window.

    In American college football, a significant part of a team’s power ranking is strength of schedule. But can you (accurately) statistically quantify a strong team in a weak conference? Some say yes, but can you really? Strength of schedule will be off. And if it’s a team like Boise State, who can’t buy a game against a major college, what of their strength of schedule? The numbers, hence the rankings, become skewed, because even an undefeated team from a weak conference doesn’t have the same strength of schedule, so it gets a lesser bowl game, and no reasonable shot at being overall rankings champion.

    Stats can say what might happen. So can the guy who stands at the betting window and flips a coin. Which isn’t to dismiss statistics as a way of trying to rank things. But it reminds me a little of the Isaac Asimov “Foundation” series, in which a mathematician predicted the world’s future based on mass statistical analysis.

    Then a mutant, The Mule, came along, and messed everything up. Sport is chockablock with mutants permanent and temporary: Messi, Torres, Xavi, some dude who one-times a shot from the parking lot and changes a match.

    Wonder what the SPI would do with Barca and Rubin Kazan before the two matches? And what about team and player form?

  13. Honestly, it’s much much much much better than FIFA rankings. And as much as some people in Europe might hate against anything coming from US sports, those who actually lead the game and are the greats, leaving Pep out people like Capello, Mourinho etc. actually have learned a lot from US type of methodology in sports.
    For what it’s worth, US sports, most of all baseball but also NFL, are the most scientific and most researched on so they already tried to learn from them. Capello famously visited baseball teams when he was on a coaching course paid by Milan to learn how the teams train the attention span of players, because obviously in baseball you have lots and lots and lots of times nothing happening but players need to be focused. That training kind of changed goalkeeping training.
    So I actually think it’s a positive step that someone from the US tries to come up with a statistical method for football, because maybe this will inspire someone in Europe to work on that too and make the method even better.

    1. You missed the point regarding my critic toward this specific American trend. The US impact on global warming is BAD. The US contribution in humanity development is GOOD. Both statements are interrelated if you look at it close. So there is no obligation to say:”Everything comes from US is great” just because there are actually some good things. Nor there is any reasoning to say:”Everything comes from US is bad” just because well…there are bad things as in anywhere else.

      I mentioned already that Stat can be the evolutional addition to football. And no one denies the fact that US statisticians are the most experienced when it comes to involve Sports with numbers. Credit for them. And I am aware of how far the concepts of sport management as implemented in the US are improving the traditional coaching methodologies. Not only from the pure coaching point of view, but also regarding team building, brand creation, and sport management principles in general. I am studying it, you know. What we are talking about here is THE WAY the betting feeders stat is approaching football. You can say that “Europeans” are not adapting to the new approach coming from the US and not customizing their mentality to it. And there is a truth in the statement that the US side (again without generalizing) failed to understand the new game coming from “Europe” and they are not customizing their knowledge to fit it. Instead, they are using the tools that worked on some other sports and apply it on this one. And it’s interesting that Kevin is pointing out that even in the US, it’s not really working. I don’t know. So I won’t confirm.

      Building lot of muscles is good for Boxing. It’s a need. It’s a must. It’s the only way to make it work. But it doesn’t mean that we can hire the same boxing coach to train football.

      I already mentioned where I see statistics as being useful in football. And I can list 101 ways to involve stat with football, and one way where you can’t (which is exactly the way approached till now, at least by the betting feeders statisticians).

      I am curious to know by the way (and I really don’t know) if this stat-ology approach is applied on some other games in US like Tennis, Golf, Volleyball, etc… Do they announce the results before the games are actually played?

    2. Actually, tennis is a great example, because they might have one of the purest ranking systems. Tournaments are weighted, and Grand Slams carry the most weight. Your ranking changes dependent upon your tourney results in the context of the tournament and the players that you defeat. Tennis rankings are probably the truest indicator of a player’s strength.

      But even they go funky when, for whatever reason, a small tournament has a number of ranked players sign up. And yet, because those same players face off against each other, you still get the props for beating folks. It’s a nice system.

      Golf is much the same way. It’s tournaments, players and ultimately, a ranking. You win, you accumulate points, you get a ranking.

      Problem with football is that the best teams don’t play each other enough. Tennis and golf are like a year-round Champions League. With sports that have divisions, conferences or national leagues, you have to devise some other method of quantifying and thus, ranking. But here’s my opinion:

      Who cares? The beauty of our game is the uncertainty, the craziness of it. Have we heard stories about someone getting rich by having bet on Rubin Kazan beating us? No, because nobody in their right mind did. Rankings or no, people know what teams are strong and what teams aren’t, etc.

      Who do rankings really help? Seeding for tournaments should be based purely on performance in tournament run-ins. But I bet you cash-ass money that Argentina is the top seed in its World Cup group. So what’s the point? Drop the ball and let ’em play.

      To Ramzi’s point, American college football is a statistical anomaly. The top team is the winner of a particular bowl game that has been dubbed the National Championship game, based on the quality of the opponents facing off in it. Fair deal, right?

      Wrong. And it’s why the system doesn’t work. There is a coaches poll, a writers poll and the computerized deal. But they all rank based on record, and strength of schedule. Yet dubious perennials such as Notre Dame always get to a bowl game, usually over a more “deserving” school. And people point out Notre Dame’s strength of schedule as a factor, as in “Well, they’re 5-4, but they played 6 teams ranked in the Top 10.” So what? They still lost to four of ’em.

      It’s why my point about power rankings is messed up with a strong team in a weak conference. We don’t really know. To keep it in this context, I counted both Rubin Kazan matches as wins. Shows how much I know. Perhaps it’s why the guys at Boise State aren’t getting the matchups against top teams that they want.

      Fun discussion, btw.

    3. That isn’t really my point. I don’t know how much you know about the Per Diem rankings from Hollinger on ESPN, just as an example, but those rankings use a completely different methodology to determine the importance of players on teams. Similar stuff the way as an example Houston Rockets assemble a team, or Moneyball etc.
      The point is not about trying to find a number and sum the game up, the point is about putting more actual science in football and progressing, and I believe every effort kind of helps, even the SPI which obviously has flaws.

      Btw. I live in Europe, so I may just praise the fact that the good old US is helping out a little bit. A major problem in Europe is that many scientists are partly savants and can’t step out of their borders for real research on fields that seem not academic. It’s equally funny that one of the first researches on spot kicks was done at an American University involving Levitt(

      And as said, they can use certain parameters from US sports and try to measure the number or whatever, but it’s not really a big deal. If it doesn’t work, they may as well work on it to make it better. That’s the whole point of improvement. So I really don’t understand what should be wrong/bad about this at all?

  14. It’s a Fun discussion no doubt. As long as no one take it as a place to scratch anyone’s national pride. My critic for this stat approach is pure football-related. I raised clear questions to indicate that this system was not, is not, and will not be valid for THIS sport. Briefing it for the last time:

    Past: What was the SPI Analysis before The last euro? If Greece was the third best, fourth best, fifth best, then we have to agree that the error margin is not a reason to damage the logic. But if Greece was the worst, second worst, third worst then its not just an error. If we took any previous tournament and applied the system, will the system give predictions that meet the actual results more than traditional football analysis (these five teams are the best, then these 4, then,,,). Will the margin of error decline enough in such analysis to worth the entire headache? That’s Past.

    Present: What are they saying? Brazil and Spain are the strongest but they are more likely to be beaten against small teams than Italy. So if Spain won the world cup “See! we told you!”, if they got knocked out:”Hey where are you all the ones who said Spain will win. We told you they are most likely going to be knocked out against small team” I mean…Isn’t that a horoscope approach?

    Future: Are there any-ANY- possibilities that this system will be perfected in predicting football results? Keep in mind that teams adventure in major tournaments are related not only to win or lose. Goal difference (Goals scored and goals conceded), head to head counters, best offense, etc…all have an impact on the teams journey in any competition. So this system has to predict the results of each and every game to get accurate, not only to say this team crosses quality is 2.5657 times better than that. If that goal is achieved, then the game is finished. Who will be interested to watch a movie after he knows the end already? So it’s an attempt to analyze football to death? If perfection is not going to happen (and I believe in that), then what’s the objective behind all this if the predictions are may be right, may be wrong? Will that differ a lot (or even match) the predictability based on normal football analysis?

    Which leads to the key question: Isn’t it better to put the effort where stat shows more potentials than showing up stat muscles? May be the realistic productive approach is not as sexy nor it water a genius ego, but its doing the right thing rather than doing things right. I wonder if the footballers are not willing to get rid of their romantic passions toward the game, or the statisticians are the ones who are romantically charmed by the overestimated magic of numbers.

    I am aware I am repeating myself, but just so we don’t keep running in circles. I quoted specific points pending for a respond.

    1. May be the realistic productive approach is not as sexy nor it water a genius ego, but its doing the right thing rather than doing things Bright.

    2. Heh, that’s a good point, Ramzi, about the “See, Spain won, we told you” or “See, Spain lost, we told you” approach. Totally horoscope style. “Today might be your lucky day, pay attention to the tea leaves.”

    3. but the point is not about perfectly predicting results. the point is about measure and improvement. perfectly predicting results is impossible. this “tool” opens up a whole new room for other people to research on a more scientific approach to football. and like I said, for what it’s worth, it’s not worse than FIFA rankings.

  15. “What was the SPI Analysis before The last euro?…”
    The SPI was not a published system at that time. You would have to take the algorithm and go backwards to verify it. What you have to remember is that statistics reveals probabilities of future events – nothing more nothing less.

    Present: What are they saying? Brazil and Spain are the strongest but they are more likely to be beaten against small teams than Italy. So if Spain won the world cup “See! we told you!”, if they got knocked out:”Hey where are you all the ones who said Spain will win. We told you they are most likely going to be knocked out against small team” I mean…Isn’t that a horoscope approach?”

    I’m sorry but I’m not sure what that I follow what you’re saying. I might have missed where the SPI Rankings state that Spain loses to small teams. Their DEF rating is actually quite respectable.

    Statistical models give you probabilities of outcomes. Yes it predicts that a team will win, and draw, and lose. It gives you an estimate for all eventualities.

    “Future: Are there any-ANY- possibilities that this system will be perfected in predicting football results?”
    Perfected as in the winner will be known 100% of the time? Such a thing is almost certainly impossible. Nor would we want to create such a system. If the beautiful game could be reduced to a beautiful number then it would completely rob the sport of any joy and suspense.

    “Keep in mind that teams adventure in major tournaments are related not only to win or lose. Goal difference (Goals scored and goals conceded), head to head counters, best offense, etc…all have an impact on the teams journey in any competition.”
    As far as I could tell by having skimmed over some of the methodology, much of this is indeed accounted for.

    “If perfection is not going to happen (and I believe in that), then what’s the objective behind all this if the predictions are may be right, may be wrong? Will that differ a lot (or even match) the predictability based on normal football analysis?”

    Ah yes, the old question of what’s the point. Indeed, what is the point of predicting the future when you could be wrong? Well, from a fan point of view it gives you a topic of discussion. It can also be really useful for finding fantasy players! From a professional point of view though, the models can provide much more than simply predicting a winner. Player ratings could indicate who is the best deal for the amount of money that they cost. They could also indicate which players you should play against certain teams. Statistics weed out players who have simply been on a tear lately by averaging out performance over a longer period of time. Really the possibilities are only limited by yourselves.

    “Isn’t it better to put the effort where stat shows more potentials than showing up stat muscles? May be the realistic productive approach is not as sexy nor it water a genius ego, but its doing the right thing rather than doing things bright.”

    Again, I’m not sure that I understand you, but if you’re saying he is simply showing off how smart he is or something then I think you’re viewing this the wrong way. When one is as smart as that guy you don’t need to go around showing it off trust me.

    Some stuff from Kevin:
    “Plus statistics don’t take into account the true variable, a single playe having the match of his life, or a defender cocking up a clearance and his side scoring a goal that completely changes a match.”

    Yes it does. Statistical anomalies exist and that’s why you rarely see 0% chance of something happening in complete models. I bet the model would not have given a high estimation for Alcorcon beating the EE. I think way more people thought it would happen than a statistical model would ever think is feasible. At the same time I think every single one of those people who thought that the EE COULD lose to them would also say that they SHOULDNT lose to them – which is EXACTLY what a statistical model would say as well.

  16. Although I love reading sports statistics and refrain from counting snowflakes, I probably agree with that comparison. I watch the games, read the stats, the papers, the player psychoanalysis, the portents, and the meaty posts and comments here–or other blogs about other teams, and then I still can’t predict most things (plus I don’t like to because its bad luck 😉 ). Although I did pick Greece in ’04, for no reason WHATSOEVER. And I’ve done as well as anyone (except my mother-in-law who is a sage) in my NCAA Basketball pools by going on most poignant sob-stories alone.

    But most importantly, if Nate was doing baseball stats in the late ’70’s, he would have missed the fact that Lou Piniella dragged the Yankees into the play-offs through sheer force of will, which as mentioned no stats would show, and probably a lot of swearing, also under-valuated. And if he was influenced by his stats, Nate certainly would have turned the lights off several times before the Red Sox won each round in 2004.

    Plus, I believe in a team from its star player to its goalie that never gets a game or its injured veteran. We can never know how much players with no stats at all contributes.

  17. Figured you’d get around to writing about the SPI. My thoughts:

    – I think maybe Kxevin mentioned this, but there is nowhere close to a statistically significant number of games to draw from. I think that’s why he’s trying to graft club data in, but the inherent noise in soccer (lots of interacting players in a fluid, contingent game) and the strong effect of tactics and match ups make the small sample size a huge problem.

    – Not sure I like that they are using club data, or how they’re using it. It weighs certain leagues much more, and doesn’t take into account how different certain player’s roles are for club and country (ahem, Messi).

    – Some of the rankings match my gut reaction (good rankings for Ivory Coast, Chile). I wonder if some of the teams he points at as underachieving (Argentina, France) are more a function of coaching than statistical variance.

    – Is the SPI meant to be predictive? Are they going to take the ratings and give match and tournament win probabilities?

  18. Great site! I absolutely LOVE the content on here. Be sure to keep up the good work and update regularly, I will check back quite often. Very informative!

Comments are closed.