Is talent identification even possible?
The dream of being able to identify sport talent in youngsters has been around for quite a while. Finding future stars based on various performance tests, mysterious algorithms, or biometric measurements seems like a dark art practiced by those who 'know' something the rest of us don't. The truth is that no real scientist thinks that future talent can be identified at an early age; there are simply too many variables involved to predict such an outcome.
Talent identification is the process of identifying future ability on an individual basis rather than a statistical one. In other words, from a group of, say twenty 11-year-olds, talent ID gurus would be able to identify the one or two individuals who will eventually become high performers.
Statistically we can say with high certainty that at least one elite athlete would emerge from this group. The difference between talent identification and relying on math is that we wouldn't know which 11-year-old this would be until it happened.
The talent ID path is based on testing, identifying the one or two superstars, and then training them for future fame and fortune. The statistical path involves training the entire group and hoping for the best. The statistical path may not sound promising but I've written several articles explaining why it is the best strategy.
So one has to ask, is talent identification as it is defined here even possible?
The simple answer is no, at least not in any practical sense. Measurement and testing are not really predictive of anything. Formulas that project growth rates or full grown size, for example, give ballpark results at best, and testing that fails to consider growth rates or relative age effect in the case of selection strategies is not much more than scientific illusion.
Some may ask, so what? Who really cares if it's possible or not? If all that sport officials were doing was trying to identify future top performers 'so what?' would be a good response. Unfortunately the way the whole process plays out is that those identified as future talents are selected for further, more advanced training with better facilities, more opportunities to compete, and working with more experienced coaches.
Those not selected receive none of these benefits. Their programs are dropped or the young athletes themselves are eliminated because the perception is that they lack talent.
The technical problem with talent identification schemes is that they only consider part of the talent equation. In sport the physical component is so obvious that it's usually the only thing we see. We rarely consider the psychosocial aspects of participation and how most athletes are engaging in sports they enjoy and, most importantly, ones that are available to them.
Large talent identification programs rely on a noninvasive battery of performance tests. But the only performance attribute unlikely to change over time is speed since it's directly related to the number of fast-twitch muscle fibers one has. Other performance attributes are trainable (endurance, strength, etc.), which makes their testing for predictive purposes essentially meaningless. Making talent ID judgements at an early age based on trainable attributes is foolish.
Anthropomorphic measurements are predictive but not of anything to do with talent. Their value lies in determining growth rates and adjusting training appropriately based on changing growth velocities.
Another use for physical measurement is to calculate ratios between body parts. In The Sports Gene, David Epstein describes how the ratio between wingspan and height is an important measurement in basketball. However, a 'good' ratio does not indicate basketball talent, merely an advantage one player may have over another. Just like a taller player might have an advantage over a shorter one.
On the true scientific front research is getting closer and closer to identifying what we think may be genetic indicators of future physical ability. But this testing will be expensive and invasive; and so far its effectiveness lies on the "maybe" side of the equation.
But even if we crack the genetic code of physical performance it will be many years before it can be implemented in any meaningful way in sports. If that happens we will then be faced with a much larger question about the future of sport and its role in society. For now the dream of talent ID is still a dream.
Bill Price (firstname.lastname@example.org) is the owner and Chief Data Scientist at Sportkid Metrics.