Why most young football talents are being wasted! On the Dutch U-17 squad and the lost generation…

Careful examination of the opening title of this post reveals that is does not intend to call the current Dutch Under-17 squad a lost generation. Admittedly, they showed a disappointing performance in the current Under-17 World Cup, contested in Mexico. After an unexpected loss against Congo, a match during which the players’ hotel was robbed too, the young side never fully recovered. A draw against a very defensive North Korean side left a glimmer of hope of qualification from the group stages, but the 3-2 loss against home nation Mexico meant elimination at the group stages for the recently crowned European Champions.


The lost generation

No, this post focuses on the true lost generation: the vast amount of players that had the inherent talent to succeed in professional football at an early age, but did not make the selection for this under-17 team. Players that did so not by their own lack of talent, but by the workings of a biased selection system throughout the youth academies from an early age on. And this bias, as will become clear later on in this article, does not seem to be limited to our small nation, but rather a worldwide neglect of potential football talent.


The Matthew effect

The current article is triggered by the first chapter of a  brilliant book, called ‘Outliers’ by Malcolm Gladwell. In this book the author focuses on how we tend to look at people around us that are capable of things that others are not: outliers. These outliers could be either talented musicians, academics, business men, lawyers, or indeed, world class athletes. Regarding the latter category, a truly insightful piece of information is derived from a simple study of the birth dates of Canadian junior hockey players.

The vast majority of these talented youngsters is shown to be born in the early months of the year, signifying a selection bias towards these players somewhere during the process of developing these young talents. Otherwise you’d have expected players’ birth dates to be roughly equally distributed among all 12 months of the year.

As an explanation of this odd distribution, the so called ‘Matthew effect’ is explained and I will provide a short summary of this theory. It implies that at an early age the pool of talented youngster is divided roughly in to two groups: those who we believe may make it on a top level and those who we believe won’t make it there. The first group is provided with extra training, extra matches, is clustered to train with better peers, hence faces better opposition and is provided with more skilled coaching. So these two group diverge during their development and an example of a player crossing over from the latter category of the deemed ‘untalented’ group to the ‘gifted’ group seems hard to find.


Talent development

So it essential for a young kid who has the inherent talent to succeed in professional football to make the cut for the ‘gifted’ group, otherwise, without the extra attention and the opportunity to work with equally gifted peers and coaches, his skills set will go to waste. And on what basis is the decision made whether or not to include a talent in the ‘gifted’ group or not? His performances on the pitch during the early stages of his development, say ages 6 to 10.

At these young ages, an age difference of six months can bear a huge influence. Being born earlier in the year provides a competitive edge over your rivals in terms of physical, neurological and mental development. And, as youth selections are based on the year a player is born, talents born in early January have rivals that are on average half a year younger, while the reverse is true for players born at the end of December. This explains why the vast majority of players in youth selections is born in the early months of the year, but it also implies a waste of the talent that is present among players born in the second half of the year.


The Dutch Under-17 squad

Teams at the FIFA Under-17 World Cup consisted of a selection of 21 players. In the Dutch squad, four players were born in January, seven in February, four in March and the remaining six in the other nine months of the year. Over 70% of players were born during the first three months of the year. In fact, only two players were born in the second half of the year: Menno Koch (July 2, 1994) and Karim Rekik (December 2, 1994).


Distribution of months of birth in the Dutch Under-17 World Cup squad


Other nations

A quick scan through the selections of the other nations confirms that this is not just a Dutch phenomenon. The average amount of players born in the first half of the year for European nations was 76%. For Australia, South America, North America and Asia, these figures were 74%, 70%, 68% and 64% respectively. The average fraction of players born during the first three months of the year for all these continents was above 40%. In Africa a reverse pattern seemed to hold true with on average over 40% of players born in the final three months of the year.


Distribution of months of birth in the Under-17 World Cup per continent


So what do these patterns signify?

The odd distribution of birth dates among players selected for the recent Under-17 World Cup implies an impressive waste of inherent football talents. With the assumptions that inherently talented players are roughly conceived year round, these data show that only a fraction of them succeed in making it to the top and these generally are the players born in the first few months of the year.

Interestingly, there were some nations with either the reverse pattern of more players born in the second half of the year (Ivory Coast 71%, Rwanda 86%) and some nations with roughly equal distribution of birth dates (Congo, Jamaica, Panama, Burkina Faso). This implies a different selection process compared to the other nations, where at least for the Dutch situation I can confirm the fact that players are grouped according to their year of birth.


A potential competitive edge

Would any nation recognize this odd pattern and succeed in transforming their youth coaching and selection system to an approach that eliminates the advantage given to players based on the months in which they were born? This would mean an enormous competitive advantage compared to rival nations that stick to this system and in fact are fishing for talent in a very limited part of the pool available to them.

15 thoughts on “Why most young football talents are being wasted! On the Dutch U-17 squad and the lost generation…

  1. basicfootball

    Great topic! I’ve read Outliers as well and it struck me how biased the system.

    However, part of the sport is physicality and it’s not often that a child is physically stronger than his peer who is 360 days older.

    It’ll be interesting to see if smaller players (Sneijder, Xavi, Iniesta, Messi) already made an impact at U15/U17 level.

  2. Orange14

    I’ve long felt that too much emphasis is placed on success on these kinds of teams. On one to the Dutch Football forums I posed the question about how successful members of the two Oranje Euro U-21 champions teams have become. The answer is not very. Other than Huntelaar, can one point to any other who is having success for club and country (and Drenthe is not a correct answer)? A bigger problem would be if late year birth players are dropping out of club academies at a disproportionate rate and I don’t know if there are any statistics for that. Often times success at these junior levels are a result of faster physical maturity rather than ultimate quality of play.

    1. Orange14

      I just looked at the Ajax full squad roster for the past season (includes a number of players who saw little or no action but were listed by the club). Results are: Jan-2; Feb-7; Mar-7; Apr-5; May-4; June-2; Jul-2; Aug-0; Sep-2; Oct-1; Nov-4; Dec-2 It looks like the trend holds up for this group. I’ll let fans of other clubs do their own analysis. Might be good if we can pool all these results.

      1. Jasper

        Last season’s squad list on the official website of De Graafschap is messed up, so I used the list on the Voetbal International website:

        De Graafschap 2010/2011, taking all listed players into account, whether they played or not: Jan: 4, Feb: 3, Mar: 1, Apr: 1, May: 4, Jun: 4, Jul: 0, Aug: 2, Sep: 3, Oct: 5, Nov: 2, Dec: 1

        Adding them up: Jan-Jun: 17 and Jul-Dec: 13. Surprisingly, only 56,7% of the players are born in the first 6 months of the year. Since it’s only a small number of players, I’m not sure what this means, if it means anything at all. But it might be interesting to check more lower ranked teams to see what effect cherry picking the talent pool by the top teams has on the other clubs.

        By pooling the three (Dutch U17 + Ajax + De Graafschap) together, we can check if there’s a significant difference between the number of players born in Jan-Jun and Jul-Dec.

        Jan-Jun: 19 + 27 + 17 = 63
        Jul-Dec: 2 + 11 + 13 = 26

        If there is no bias, a player should have a 50% chance of being born in the first half of the year. Using R commander, we find that the difference between Jan-Jun and Jul-Dec is significant (p binom.test(63, 89, p=0.5)

        Exact binomial test

        data: 63 and 89
        number of successes = 63, number of trials = 89, p-value = 0.0001099
        alternative hypothesis: true probability of success is not equal to 0.5
        95 percent confidence interval:
        0.6019310 0.7994924
        sample estimates:
        probability of success

      2. Jasper

        Didn’t know I’m not allowed to use less-than and greater-than symbols, so that last bit of my comment looks a bit messed up. Trying again:

        If there is no bias, a player should have a 50% chance of being born in Jan-Jun. Using R commander, we find that the difference is significant (p less-than 0.01).

        binom.test(63, 89, p=0.5)

        Exact binomial test

        data: 63 and 89
        number of successes = 63, number of trials = 89, p-value = 0.0001099
        alternative hypothesis: true probability of success is not equal to 0.5
        95 percent confidence interval:
        0.6019310 0.7994924
        sample estimates:
        probability of success

  3. kv

    Good article, but I am a bit skeptical about the theory proposed by Gladwell. I don’t think a period of 6 months can make a big difference in physique so as to give an advantage to those elder by 6 months! The elder player cannot be 1 foot taller and run twice as fast as the player who is just six months younger. I myself studied for four years in a class where everyone were on an average one and a half years older than me(some were even 2 years older than me) and the others weren’t too different to me.

    1. 11tegen11 Post author

      Although I welcome all sorts of scepsis, I disagree with you on the fact that 6 months of age difference won’t make a huge difference at a young age. Remember that this age difference only comes into play in dividing talented players of 6 or 7 years into those who are deemed to make it at the highest level and those who are deemed not to make it there. From that moment on, this separation becomes a self-fulfilling prophecy as the first group is supplied with more training hours, better quality opposition and better skilled coaches. The superior nurture of the group that was judged to make it at the highest level will be responsible for a diversion in the level of play between both groups of players.
      And at very early ages a difference of 6 months might have a huge impact on physiological, neurological and mental parameters that determine the actual level of play at that very young age.

  4. D.

    This is a very interesting point. I think the smallest of age differences can make a difference at very young ages. In fact, I’m quite sure Feyenoord have already implemented a system to nullify this effect in their youth academy. This was probably not implemented in time to make an impact on this generation of players, seeing as there are Jan-Feb born Feyenoord youth players in the U17 team in abundance. I can’t find any data on Feyenoord’s new birth date policy, but I think I read in Volkskrant or NRC (for the Dutch readers) somewhere.

    I however do not agree entirely that this early shift is all that crucial. Really, looking at some of the world’s top players, very often they only joined a top club/academy later than the 6-10 years of age period. Messi, Iniesta, Pedro and Busquets were all 13 or older when they joined Barcelona, and players Huntelaar, Van Persie and Van Nistelrooy only went to a top flight club quite late in their teenage years (or in RvN’s case, not until they were well past 20). They seemed to have fared quite well, haven’t they?

    Not to play down your argument too much – I really think youth academies should focus on this aspect a lot more -, from the players’ point of view it’s not quite so disastrous to ‘miss the boat’ when they’re 6 years old.

  5. Jasper

    When you look at the full squad lists of the teams that qualified for the 2010 World Cup, you can see that the trend is still there at adulthood at the highest level. Although the gap has narrowed considerably, there is still a significant difference between the number of players born in Jan-Jun and Jul-Dec (exact binomial test, with 0.5 probability of success and 95% confidence level).
    When you examine the data per continental confederation, it turns out the differences are no longer significant, except for CONCACAF, AFC & OFC (which were pooled because they contained less than 100 players each).
    But since the differences are nowhere near as spectacular are they are in U17, we can safely say that most of the players born in Jul-Dec are not so much lost, as they are taking a detour.
    Data from http://www.fifa.com/worldcup/archive/southafrica2010/teams/

    TOTAL 2010 WORLD CUP: 399 – 337
    Out of the 736 players, 54.2% was born in Jan-Jun. The difference is significant (p = 0.02448)

    UEFA: 159 – 140
    Denmark: 11 – 12
    England: 11 – 12
    France: 8 – 15
    Germany: 14 – 9
    Greece: 16 – 7
    Italy: 11 – 12
    Netherlands: 12 – 11
    Portugal: 15 – 8
    Serbia: 10 – 13
    Slovakia: 10 – 13
    Slovenia: 12 – 11
    Spain: 14 – 9
    Switzerland: 15 – 8
    Out of 299 players, 53.2% was born in Jan-Jun. No significant difference (p = 0.2979).

    CONMEBOL: 63 – 52
    Argentina: 16 – 7
    Brazil: 13 – 10
    Chile: 11 – 12
    Paraguay: 8 – 15
    Uruguay: 15 – 8
    Out of 115 players, 54.8% was born in Jan-Jun. No significant difference (p = 0.3511).

    CAF: 71 – 67
    Algeria: 13 – 10
    Cameroon: 15 – 8
    Côte d’Ivoire: 13 – 10
    Ghana: 9 – 14
    Nigeria: 9 -14
    South Africa: 12 – 11
    Out of the 138 players, 51.4% was born in Jan-Jun. No significant difference (p= 0.7985)

    CONCACAF, AFC & OFC: 106 – 78
    Honduras: 14 – 9
    Mexico: 16 – 7
    United States: 15 – 8
    Australia: 9 – 14
    Japan: 17 – 6
    North Korea: 7 – 16
    South Korea: 16 – 7
    New Zealand: 12 – 11
    Out of 184 players, 57.6% was born in Jan-Jun. The difference is significant (p = 0.04624)

    1. 11tegen11 Post author

      Now, how’s that for a comment? Thanks Jasper for putting in this effort and neatly sorting all these data out!

      It is interesting to see that the supposed Matthew effect seems to get diluted over time, which is something that I’d guess is to be expected with the biggest talents born in the second half of the year more or less working their way around the bias that the system imposes on them.

      There are two thoughts in my mind regarding this subject after reading your extensive analysis.

      First, it is not part of the hypothesis that the effect comes down to the first months of the year in every country. For example in my data set some African nations showed a different, yet very unequal, spread over the year. Ivory Coast and Rwanda had 71% and 85% of players born in the second half of the year. Burkina Faso had 76% of their players born either between April and June or October and December. Jamaica had an unequal spread in the contrary direction, with 67% of players born in either between January and March or between July and September.
      This goes to show that, for all I (we?) know at the moment, there might be a different selection system going on in these countries. The latter two nations might apply two age groups out of players born in the same calender year, while the aforementioned African nations might have a cut off date later in the year instead of January 1.
      It would be interesting to see if we could find out the selection dates that applied during the years that this generation of players was brought through the youth ranks in order to treat the data correctly.

      My second thought on the matter is that the effect might indeed be more pronounced in the current Under-17 generation than it is in the generation that featured during the World Cup. The amount of money invested in young football talents has grown immensely over the past ten years, so the diversion in talent development between the group of talents that were deemed ‘gifted’ at an early age and those that were deemed not to make it (the Matthew effect) might be much more pronounced in recent generations. Which would be a very worrisome trend if true…

      Again, thanks for your extensive contributions so far!

      1. Jasper

        Thanks for your kind words. You’re absolutely right that there probably are country (and regional) specific differences that have to be taken into account, but when you look at your graph ‘Distribution of months of birth in the Under-17 World Cup per continent’, it seems you’ve stumbled onto a continental and mostly global trend and a couple of comments here made me curious about how that holds up at adulthood. That’s why I decided to check the World Cup squads at a continental level, so I would be able to compare them to that graph. The country list is merely there because that’s what I used to add things up and I included it thinking there might be someone who would find it interesting. Besides that, those individual numbers are actually quite meaningless, because squads only contain 23 players and that number is just too small to base conclusions on.

        As you can see, the gap has definitely narrowed at adulthood, which, I agree, could mean a couple of things. Either most talented players born in the second part of the year beat the system and do make it eventually, or this trend is quite new. I’m guessing the trend might have become more pronounced, because it’s true that “the amount of money invested in young football talents has grown immensely”, but I don’t think it’s that new. The crucial part of deciding who’s going to the gifted group and who’s not is still mostly done the way it’s always been done: scouts using their eyes and judgement. The difference may have become bigger, but the gifted group has received far more and better training for decades. And, according to Gladwell, it is practice that makes perfect.

        What’s also important to realise is that 16 might just be too young an age to judge players, meaning that those that are successful at that age, don’t necessarily have to be when they’re adults. Take for instance one of the earliest successes of a Dutch U17 squad: coming in third at the 2005 U17 World Cup. That’s also the biggest success we ever had at a U17 World Cup, so it’s only natural to assume that this should be a golden generation. But when looking at the list of players, most of whom are 23 by now, it struck me that hardly any of them have made it big time yet (by the way, at least in this case, the trend seems to be already there: 14-6).

        As I said before, you’re right about looking at the differences in selection systems per country, but there are also a couple of problems with that:

        – The U17 selection doesn’t stand on it’s own, it’s part of a bigger process. I have never heard of a player from an amateur club being in a national team, so it’s the clubs that do the first part of the selection. Since talented youngsters are being shipped across the globe these days, this could mean that different systems are getting combined.

        – You can’t really examine national U17 squads per year, since it only contains about 21 players and that number is just too small to draw conclusions from. Going back in time and adding things up isn’t an option, because of the massive fraud that has been going on. From The Guardian: “Reports claim that retrospective analyses of the previous three Under-17 World Cups showed more than a third of all players were too old.” http://www.guardian.co.uk/football/2010/feb/21/nigerian-football-age-old-problem

        – In parts of the world, sadly, food availability might be an issue. For instance, players born in winter (June, July and August in Africa) might have a smaller chance of making it as a soccer player, since they could be underdeveloped from the start.

        The best thing to do would probably be to stick to European countries only.

  6. ABC

    Keep in mind that things work differently in certain places, e.g. England. The year groups are worked out from September to August, as opposed to from January to December.

  7. Johann Roppen

    This is a very complex issu, and to gather relevant data is a good start. But as the discussion shows, one can only get so far by looking at the data from a (small) sample of players, and it’s a sober remark to look closer at the different systems for development/selection of players in different countries.

    One kind of data is missing in the discussion, and that is demography. The number of child births will vary through a year. In my country, Norway, in 1994 a majority of children were born in the first half of the year. Each month from February to July counted a number of childbirths above the average. Thus one in general must expect a higher number of players being selected from the first half of the year.

    1. 11tegen11 Post author

      Thank you for raising this excellent point.
      This comment section really does what it’s supposed to do, raise valid points aiming to improve the original article.

      Though your point is valid and the amount of births does vary throughout the year, I hadn’t thought of incorporating that into the analysis. However, as the next table shows, the variation in the amount of children born each month is limited.

      51.5% of Swedish children is born in the first half of the year, versus 48.5% in the second half of the year. Data are from the year 2006.

      Data from the US show a different picture. Look at the graphs below.

      Despite this distribution favoring children born in the third quarter of the year, only 2 out of 21 players in the US national U-17 team were born in July, August or September and 10 of 21 (48%) during the first three months of the year. This selection bias goes against the direction of the bias induced by the amount of children born in particular months.

      So, I think your point is valid. Incorporating difference in the amount of children born throughout the year would improve the validity of the conclusion drawn, but from this early exploration I don’t believe that the distribution worldwide is that uneven (with the US data even going in the reverse direction).

  8. Pingback: Tangerine Talent Timeline | Tangerine Dreaming

Leave a Reply