It’s been said all too often that the United States produces too many athletes and too few soccer players. And you know, it really is true that native talent from the States is less technical. We don’t produce the best dribblers. We don’t produce a lot of highly creative players. And we don’t develop our players to be able to play in small spaces. Instead, we excel at players who are able to leverage their strength against the opposition, to use their height and their jump to rise above the competition. To be bigger, to run faster, and to keep going longer. The way that our national teams and our leagues play reflects this. While MLS is now stocked with talented Central and South American talent to put in that crisp through ball (players like Federico Higuain, Nicolas Lodeiro and Miguel Almiron), the overall game remains very rough and rugged just like most of the American players on the rotation.
So, what’s going on and how do we fix it?
The birthday effect problem
An explanation that many people, including myself, ascribe to at least part of a problem is that American soccer has a systemic bias towards physicality in player development. The idea is that, because American youth soccer is set up with the expectation that organized youth team results are important and that winning is important, kids who are able to dominate games at a young age end up getting more time and attention than other kids. In any age group, there will be a range where some kids are more developed than others. Some kids just develop early. Other kids are simply younger, with the difference between a January and December birthday being pretty significant, especially in young children. A third example would be kids who were born prematurely and thus have bodies that would match up with a younger child’s. Because of reasons like these, among others, some kids will be taller than the other kids they play with. Or maybe they’ll be stronger. Or faster. Or bigger. And those kinds of discrepancies can have a meaningful effect when those kids play.
It makes sense that kids who are bigger and faster are going to be able to dominate over smaller kids. Indeed, we see something similar happening in the classroom. In sports, the NCAA has seen this as well, calling it the “Birthday Effect” and tracking data for it in soccer, among other sports. The problem is that those physical traits can even out over time. Kids stop growing in their late teens and early twenties. Until that point, a child that’s more developed than his peers will have a physical advantage. But after that point, that advantage goes away, simply because everyone else will have caught up. The worry is that kids who develop early will begin to dominate games and, as a result, learn to rely on their physical capacities instead of learning technical and tactical skills. And because those kids dominate games, youth coaches may be inclined to focus their resources on those kids, since that may result in winning more games. As a result, the United States produces fewer technical players and lots of physical ones. That’s where the U.S. Soccer Federation is trying to step in with this Bio Banding initiative:
Bio Banding Explained
The idea is that, by matching players to a maturation group rather than a birth year group, kids will be able to showcase their talents in an environment where other players don’t have a physical advantage just for being more developed. They match the groups by measuring maturation. And they figure out maturation by predicting a child’s future final height (using current height, weight, and parents’ height) and comparing how close they are to actualizing that height. Then, those kids are paired with other kids who all fit within a 5% range. On paper, it’s a clever solution using data to get around the development bias problem.
That sounds good, right? Yeah, about that...
It’s Too Good to be True
In principle, this could be a good idea. But, there are massive problems with the idea of using this data to estimate a child’s maturation. To get at the heart of the manner, you don’t actually know how developed a child really is by doing these calculations. You are merely making an estimate. I don’t mean to question the mathematical modeling that they use here. Apparently, they use some well-regarded theorem, but algorithms like this are meant to be used on very large sample sizes. We are talking hundreds or thousands of data points. When you have data that large, you can really start to be able to make accurate, generalized statements. However, soccer doesn’t use large sample sizes. The sample size we are talking about for kids is the size of the club choosing to use bio banding. In the context of soccer, we aren’t making predictions for hundreds or thousands of players. We are making predictions for individuals. And on an individual level, these things can vary pretty wildly.
Let’s use myself as an example. They said that they use the parent’s height as a comparison, so let’s compare me with my dad. I’m around 5’ 8”, somewhere between 4 and 5 inches taller than my father. That makes for a 6% difference between the two of us. So, if bio banding predicted that I would be around my father’s height, I would have been assigned a score that was 6% off. That’s the difference of an entire interval! If my true maturation score is at the 85th percentile, I ought be in the 80-85% group. But the score I am given is 6% higher, I’d leapfrog past the 85-90%, to the 90-95% group. With that kind of range difference, an error like that could completely misplace me! That’s the thing with small intervals. If the statistical variation is predicted to be just +/- 2.5% (i.e. 5%), then half the kids in an interval will be misplaced. The margin of error has to be considerably smaller than the intervals, otherwise it doesn’t work. Even a margin of error of 1% will mean that 20% of the kids on the field will be mismatched. And, of course, large intervals undermine the whole point of grouping kids up based on maturity in the first place.
The video mentioned that the system also uses height and weight in conjunction with parent’s height. With a single data point, you can’t tell the difference between a kid who has finished growing and one who has a lot more development left. As a 14-year-old, I was pretty short. I was only around 5’0”. In that context, it would have made perfect sense if I only ended up at 5’4” or 5’6”. At the same time, I could have been an inch or two taller (like my brother is). From that single point at 14, you can’t tell just how much taller I’d be.
On top of that, you need to factor in different contexts. American soccer players disproportionately come from immigrant families. But there are differences in height between people who grow up in one country vs. the United States. Cristian Roldan, for example, is 5’8” and is of Salvadoran and Guatemalan decent. The average height for a man in El Salvador is 5’5” (Guatemala doesn’t seem to have a report on the heights of its men). Omar Gonzalez is 6’5”, but the average height is 5’7” for Mexicans that live in Mexico. This isn’t to say that they wouldn’t have been as tall if they had grown up in those countries. Rather, it is to make a point about how context matters. Case in point, while Omar Gonzalez is taller than the average Mexican, he is also taller than the average American man, who stands a tad above 5’9”. It’s all about height variations:
“Height variations within a population are largely genetic, but height variations between populations are mostly environmental, anthropometric history suggests. If Joe is taller than Jack, it’s probably because his parents are taller. But if the average Norwegian is taller than the average Nigerian it’s because Norwegians live healthier lives.” (Source: The New Yorker)
Because of such contexts, making individual predictions for a child’s eventual height is very tricky. And it matters because a difference of just of inches could completely misplace a child in a sample based on development. Even misplacing a few kids would have compounding effects. Let’s say that there were 100 kids who were evaluated by the Bio Banding process and put into a single developmental interval. Let’s also imagine that only 10 of those kids were improperly placed (10% of that 5% interval, so an error rate of 0.5%), with half of those misplaced kids underdeveloped and half overdeveloped. In that context, the 5 overdeveloped players will have a big advantage, especially now that all the kids they are competing with will be leased developed than they are. Those five will get more playing time and more time investment, and that means that you are going to continue rewarding physicality over other traits.
What happens to those five kids who are underdeveloped? Well, they are screwed. They’re going to see even fewer opportunities than otherwise. Especially since there’s now scientific evidence saying that these kids are underperforming in their supposed development group while the advantaged kids are overperforming. If coaches are going to be using technology like this, they can’t be expected to look at their players’ performances and then discard what that tech is telling them. That’s just not how people interpret information like this.
Problem of Philosophy
Of course, all of this is merely on top of a big philosophical issue. One particular quote from the clip in U.S. Soccer’s tweet stood out to me:
In order to improve the overall quality of soccer in the US, it’s important for coaches to have the tools to identify and cultivate the best players, not just the ones who mature the fastest.
There’s a few big issues there. “It’s important for coaches to have the tools to identify and cultivate the best players.” That right there strikes me as problematic. It assumes a few things. First, that talent is something innate, rather than something that is learned and cultivated. Second, that soccer development is a zero-sum game and that those who show promise at a young age should get the lion’s share of the investments. And, third, that soccer in this country is about the best players, or at least about playing at the highest levels. All of these are cultural critiques, and they reinforce one another. If soccer is about producing talented players who will go and win in domestic and foreign leagues and triumph at the international stage, then it is necessary to focus on only the best players. And if generating that talent is about finding those with natural talent and helping them grow into the best players possible, then youth soccer is merely an exercise in identifying those players.
But I reject all of those assumptions.
Talent isn’t merely about coaches being able to find the players who might become highly talented. It’s not all about investing in the person with the right nature. I inherently reject the idea that a player’s talent level is destiny. Instead, I think it’s much more about the nurture. Coaching needs to be about helping a player learn and grow, regardless of their skill level. A good coach will make any player better, not merely give playing time to players who are already good. The best coaches don’t merely find the best players. They make the best players. In that vein, the language here is a pretty big indictment of the quality of American coaches. It’s saying that the job of the coach is to perform basically as scouts, to find talent and give them playing time in an appropriate environment, and that our youth coaches are pretty bad at doing this limited task. The solution to this ought to be to increase the quality of our coaches. To make sure more people who understand the game go into coaching, that they get training, and that they are accessible to players.
Soccer isn’t about making sure that those who have talent get better. Soccer is supposed to be the people’s game, not the elite’s game. It’s often called the democratic game, an incredible meritocracy. But if we exclusively focus on only those who we think are the best, we are inherently going to miss out on a lot of cool talent, regardless if we are looking for physicality or technicality. Soccer, and especially youth soccer, shouldn’t be about getting the best players on the field. It should be about getting everybody on the field.
Soccer isn’t about winning Champions League or the World Cup. It’s about kicking a ball into a net. It’s play. It’s fun. It’s absurdity and chaos. It’s community and connecting with other people. It’s expressing yourself and showing what you can do, both individually, and as a team. That national culture of the game has to be more than the first division and international level. It also has to be about kids playing against each other, about adult pick-matches, and about unwatched fourth division league games.
It’s these kind of cultural assumptions that, when tied with a win-now expectation, has stifled American talent. And we should reject that kind of language.
How the rest of the world does it
In any case, the example from the rest of the world indicates that this isn’t the way to go. Kids in other countries don’t grow up subdivided into percentiles of maturation. Yet players from Italy to Argentina to Japan to Mexico to Nigeria end up producing more technically astute players than us. Indeed, the example in many countries is to divide players less. What happens when you have kids just playing the street together? They naturally have age variances. Some kids will be 5 or 6 years old. Others will be 8 or 9. or maybe they’re teenagers, ranging from 13 or 14 to young men in their early 20’s. And they play together. Those younger kids also have to deal with players who physically dominate over them. But the difference is that everyone in their age group experiences that. All of them have to learn how to play more technically and with positional intelligence in order to play well. But over time, the older kids age out. They may stop playing or move to a club team or simply play with older kids. The small young kids become the older kids. But they’ve still gotten the chance to play and develop those skills.
In the United States, we have exactly the opposite culture. We don’t play pick-up matches. We play organized league games. Four year olds are grouped up with four year olds and chase each other on a field. Ten year olds play organized league games with ten year olds. We field U-14 and U-15 and U-16 teams. High school students are split between the older kid in the varsity team, while the younger ones play in JV team.
That’s the real place where we have to change things. We have to change how we play, and we have to change our soccer culture. We have to change the things that incentivize and reward physicality over technical ability, positional intelligence and teamwork. And we have to do this holistically if we want things to get better, not these misguided attempts to subdivide kids even further.