"That kid has talent" is something you hear a lot at the range, said on the strength of a few ends. Research on archery lets us say something more precise about what can be measured and how far it can be trusted. The simplest signal, an expert coach's judgement, is linked to score, but in the only study that has set it against levels it picked out the archers who were behind and did not rank those who were already ahead. Physical tests, attention tests and computer-built models are linked to score more weakly and less consistently than people think, almost always in small samples.
All of these findings share one limitation, and it is the one to keep in mind when choosing between athletes: they are snapshots taken at a single moment. They tell you who shoots better today, not who will shoot better in five years, because no study has followed tested youngsters over time to see who reached the top. Assessment batteries remain useful tools for those who work with groups, coaches, clubs and federations, which are the settings I work in, provided we know what they actually measure.
What talent means
In sport, talent is the set of predispositions that make development faster and raise its ceiling. In archery it has physical components (strength, endurance, build), control components (postural stability and fine coordination), and visual and mental ones, from attention and handling competition to the capacity for learning the movement.
Because there are so many components, no single number can describe them: an athlete can be very strong physically and fragile emotionally, or the other way round. That is why a rigorous battery, rather than giving a mark out of ten, returns a profile like the one in the figure, with strengths and vulnerabilities. A profile, however, describes the archer of today, and how much it says about their future is the question the studies answer least.
The coach's judgement
Before any test there is the coach's eye. Two studies of archers have set technique rated by experts against score, and both find a link: in 62 male recurve archers at the Indonesian national championship, technical quality is the factor most closely linked to the FITA round score among those measured, and in 31 Thai archers with international experience a ten-item technique sheet is moderately linked to score (r around 0.4). Both have limitations that matter: the first does not describe how technique was rated, the second appeared in a journal with questionable editorial standards, and neither describes a rating made without knowing the archers' scores.
The most useful finding for anyone who has to select comes from a Swedish study from 1999, designed for a quite different purpose. Four national federation coaches split twenty archers aged 13 to 18 into three levels, rating their overall ability rather than technique alone, and they agreed closely with one another (multiple correlation of 0.90). Over the twelve arrows at 18 metres of the control condition, the group rated weakest scored 91.3 out of 120, the intermediate group 105.6 and the elite group 105.5: the coaches' rating separated the archers who were behind, while intermediate and elite scored the same.
It is a hint drawn from twenty young archers, with only twelve arrows and three kinds of bow mixed together, and the weakest group were not beginners but good club archers. It does, however, point the same way as other findings we will come to, starting with reaction time: several measures separate those who are starting out from those who are ahead quite well, and archers who are already good from one another much less well, which is exactly the part that matters for selection.
Throwing tests and reaction time
Among beginners, researchers have looked for something outside shooting that anticipates the score. In 17 secondary school pupils at the end of a basic course, those more accurate at throwing a tennis ball at a target on a wall shot better with the bow (r = 0.51), but in the same study none of the other eight measures, from height and arm length to shoulder position sense, was linked to score, and with nine correlations a single one above the threshold may simply be chance. As early as 1969, in a course for beginner schoolgirls, a general motor ability test was only weakly linked to the final score (r = 0.32).
Reaction time follows the same pattern as the coaches' rating. Between the clicker snap and the release, experienced archers are clearly quicker than beginners: about 100 milliseconds against 200 in one study using electromyography, and 112 in the elite, 140 in the intermediates and 167 in the beginners in another. Among archers who are already good the difference disappears: in seven archers from the German national squads the mean time was not linked to score, and in fifteen French archers it did not separate the elite from the intermediates. Measured away from shooting, in 127 American archers, reaction time was only weakly linked to score, and in the comparison between groups those scoring above 1,200 reacted in 181 milliseconds against 188. On the consistency of that time the studies are divided: in the seven German archers the most consistent scored most, in fifteen young Ukrainian archers they did not. That better archers have slightly shorter and above all more consistent reaction times is therefore contested.
Physical tests and predictive models
Physical tests are the easiest to administer and the most common in youth batteries. There is a link with level, but it is more limited than people think and is often tangled up with body size and age.
Grip strength is the clearest example. In comparisons between levels, better archers have a stronger grip: in twelve young Malaysian recurve archers split in half by score, the better six gripped 35.2 against 23.8 kg with the left hand and 33.8 against 29.3 kg with the right, but they were also twelve centimetres taller, and height and arm span were more closely linked to score than grip was. Within a homogeneous group the link disappears. On arm strength in general the studies contradict one another: grip measured with a dynamometer is not linked to score in a long series of samples and is in others, and the highest correlations come from small groups that mix beginners and experienced archers, where any physical measure ends up separating one from the other.
The other qualities have limited support too. The vertical jump has been linked to score in a single study, of fifty young Malaysian archers, with a shooting test of six arrows, and in the same study the standing broad jump, which measures a closely related quality, stayed below the threshold. Single-leg static balance is not linked to score in seven samples out of nine, while body sway while aiming, measured on a force plate, nearly always goes with better shots; it is, however, a measure taken during the shot, not a separate test. As for isometric holding, the two studies that come closest to the question do not find it more important than maximal strength.
Machine learning models have also been built on these tests. A series of Malaysian studies from the same research group splits young archers into high and low potential with accuracies between 82.5 and 97.5%, a figure that impresses until you look at how the groups are formed: they are built from the same measures and the same score that the model then has to recognise, so the accuracy is largely circular. The papers most probably rest on a single sample of fifty archers analysed several times over, the models go astray mainly on the weaker archers, and the only test on archers kept out of the training data involves ten youngsters, one of whom was misclassified. That a battery of physical tests identifies the most promising youngsters remains a reported finding, but the evidence concerns current level.
The batteries built with more care, validated by the judgement of seven experts or repeated a day later, show that the instrument is consistent and repeatable, not that it picks the right youngsters. And a battery used on forty children at three Indonesian clubs placed nobody below the middle band, so it did not discriminate downwards. A good result in physical tests therefore describes the starting point, not how far a youngster will develop with method and consistency.
Attention, mind and vision
On the mental side the answer depends heavily on the instrument. When archers are asked how concentrated they feel, as in a 1986 study of over a hundred archers up to world level, self-rated concentration does not go with score (r = 0.16). When an attention test done against the clock is used, the sign is nearly always the same: in sixty-one archers from the Chinese national squad a pencil-and-paper test was linked to performance in the following week's competition (correlation 0.43), and three small Indonesian studies from 2026 find values between 0.62 and 0.79. Here too the picture cracks, because in forty-five traditional Turkish archers only two indices of the d2 test are linked to score, and in one of the Indonesian studies the link disappears once arm strength is taken into account: that is why the link between attention and score remains contested.
How much mental factors weigh overall has been estimated several times. In 1986, in 79 American recurve archers, seven measures accounted for 58% of the score differences, and the mental block (dwelling on mistakes, confidence, visual imagery) was worth 26% of it, almost twice strength and build together; widening the sample to 124 archers, however, brought the adjusted share down from 53 to 39%. In sixty-one Chinese national squad athletes, psychological factors account for about a third of the variability in results. These are shares of association: they say that a sizeable share of the score differences goes together with mental measures, not how many points would be gained by changing them.
On the mental toughness of medal winners there is a single direct comparison, of forty Indian university archers: the twenty medallists reported far more toughness than the others, with 4.45 weakness points out of 24 against 14.20. The study appeared in a journal with questionable editorial standards, the questionnaire was filled in once the medals had already been awarded, and winning may make people more confident rather than the other way round. The registry reports it but does not use it to decide the status, so the question of medallists' mental toughness remains not studied.
Vision, finally, is often measured and counts for less than people think. In the only study that has set visual acuity against score, of a hundred and two recurve archers, none of the eleven visual measures was linked to the season's average score. On cross dominance between eye and hand the studies are divided: in 82 sport science students picking up a bow for the first time, shooting without a sight, those with dominant eye and hand on opposite sides scored fewer points, while among trained archers the advantage does not show.
Years of practice, starting age and starting point
A test given to a youngster also measures, without saying so, how long they have been shooting and how old they are. The studies that set years of practice against level point in opposite directions: in the Chinese national squad, international-level athletes had been shooting for ten to thirteen years and first-class athletes for six to eight, while among 62 archers at a German championship and a European one, years did not clearly separate the two levels (11.8 against 15.7, p = 0.061) and arrows shot per week did (273.7 against 696.5). Whether years tell how good an archer is is therefore contested, and so is the opposite reading, that on their own they do not tell an archer's level. In these comparisons years and age rise together, and none of them counts the arrows shot in those years.
On starting age the data are even more indirect. No study has asked the question, and the data come from sample descriptions: high-level Korean archers had started at around ten, while the eight international masters on the Chinese squad preparing for Tokyo 2020 had 4.9 years of practice at 19.6 years of age, so they had started at around fifteen (the Chinese figure, though, comes from a questionable venue). Starting age seems to depend on the country more than on the level, and nobody has counted how many children started early and never reached the top: that archers who reach the top started as children remains a reported finding, read from samples described for other purposes.
Then there are the archers who start further back. No study of archers was designed to tell whether those who start lower gain more points, and the small programmes that hint at it point in different directions: in twelve youngsters at a club, after eighteen circuit sessions, the archer who started on 204 points gained 77 and the one who started on 226 gained 41, while in two other programmes the gaps between archers stayed the same or widened. None of these studies has a control group, and an archer who shoots very badly on the day of the first test tends to shoot better the next time even without training.
Taken together, these findings describe the easiest mistake to make when selecting on a test. Whoever is ahead today may simply be older, taller or have been shooting for longer, and a test taken at a single moment rewards that without saying anything about how far a later starter will go.
Predicting the score
If the aim is to know how an archer will shoot, their past results are the natural place to start, and the studies only partly bear this out. In a machine learning model built on thirty-two Indian archers, trained on the arrows of one session and tested on the next, the error was about six points out of three hundred; at the top of the predictors were years of practice, in first place, and scores in previous competitions, in third, but in second place was resting systolic blood pressure taken that day. In an indoor competition among seventeen adolescents, shooting time and heart rate variability before each arrow accounted for 11.8% of the score of the single arrow, while the difference between novices and experienced archers, on its own, weighed about twice as much.
In small groups the internal ranking stays stable even when conditions change a great deal: the correlation is 0.898 between the scores of nine young archers before and during the pandemic, that is, across two very different periods of training, and 0.985 between the start and the end of six weeks in ten pupils. Nobody, though, has yet compared a model with the simplest forecast, the average of the last few competitions, and the two studies that set moment-to-moment data against past results do not give a clear answer: that the score is predicted better by experience and past results is contested.
How many trials are needed
A test is only worth something if, repeated under similar conditions, it gives a similar result. You often hear that one trial is not enough and that measurements on archers must always be repeated, but the answer depends on what is being measured. For maximal draw force, as early as 1967 three trials in the same session on 49 university beginners gave very close values, with a correlation of 0.91 to 0.98 between the best trial and the mean of the other two: a single trial ranked the archers almost as well as three, which is why the idea that one trial is not enough is contradicted.
For the electrical activity of muscles things change. In twenty experienced archers, an electromyographic measurement taken once had a reliability of 0.745 and the mean of five trials 0.936, but the signal fell trial after trial, from 5.87 to 4.64 mV, and the hip angle at which it was measured weighed almost four times as much as the differences between people. The study appeared in a journal with questionable editorial standards, and it covers a single muscle. Repeating helps when the signal is noisy, as long as there is time to recover and the measurement is always taken in the same position.
The same reasoning applies to the shooting test that ends almost every battery. The link between jump and score rested on six arrows, the difference between the Swedish coaches' groups on twelve, and with so few arrows two archers who are close together cannot be told apart. A serious protocol fixes the number of arrows, the distance and the conditions in advance, and that is what makes it possible to compare a result over time rather than rely on an impression.
What it is for, and what it is not
For all these reasons, talent assessment is not a tool for weeding people out. The links I have described are associations collected at a single moment, and judging a youngster's development on a test at thirteen asks those data for something they do not say. What it is for is understanding where an athlete is strong and where they are vulnerable, and therefore what to work on and how to allocate the resources of a club or a squad, which are always limited.
Used in this way, a battery turns vague judgements such as "has talent" or "not cut out for it" into a profile to start from, and repeated with the same protocol it shows whether the work is paying off.
ARCOS: from scattered tests to a shared archive
Everything I have described also has a practical limitation: a test done properly at one club and one done however another club sees fit cannot be compared, and the numbers stay in the drawer of whoever collected them. The question that matters most for selection, whether a youngster who does well on a test at thirteen goes on to reach the top, can only be tackled by following the same archers for years with the same measures.
That is where ARCOS, the system I am working on, comes from. It starts from the competition results that already exist and links them to each archer's membership number, so that every archer has a history, and adds training diaries and assessments collected with the same protocols for everyone. It is a performance intelligence system, not a medical tool: it reads performance and load over time and offers technical readings and corrections to support coaches, with no diagnosis and no selection on health grounds.
Adopted at federation level, a tool like this would make possible things no single club can do. With data from many archers collected in the same way, the same youngsters could be followed for years to see which measures taken when young go with the level reached as adults, in Italian archers; the competition history would give each archer the simplest reference for predicting the following season; and coach education and squad preparation could start from comparable numbers rather than impressions. It is a project still in development, and it needs to be tested in the field before it becomes a promise.