Does sport-specific cognition predict a player's future level?

Sport-specific cognition does predict future playing level, with a moderate to large effect that is real but not decisive. Murr et al. (2018) found decision-making predicts later success with d = 0.81, one of the three strongest predictors in the prospective literature. Ivarsson et al. (2020) describe psychological factors more cautiously as small but statistically significant. The largest prospective study in Europe, Höner et al. (2021) with 13,869 players, shows that no single measure beats a combination, and Höner et al. (2023) set the reliability bar a sport-specific test should meet.

Present level versus future level

Talent identification asks two different questions. “Who plays better now?” can be answered with a cross-sectional study: test everyone once, compare groups. “Who will play better in three to seven years?” needs a prospective design: test now, wait, then check who succeeded. The second kind of evidence is rarer, harder and more valuable, and it is the only kind that justifies using a test for selection or development decisions.

Murr et al. 2018: the prospective review

Murr and colleagues, with Höner among the authors, reviewed every youth football talent study from 2000 to 2016 that used a prospective design. Effects in Cohen’s d:

Predictor	Effect size (Cohen’s d)	Type
Ball control	0.57 to 1.28	Psychomotor
Dribbling	0.47 to 1.24	Psychomotor, the most studied
Decision-making	0.81	Perceptual-cognitive
Hope for success (motivation)	0.27 to 0.74	Personality
Fear of failure	0.21 to 0.30	Personality

Decision-making is the only perceptual-cognitive factor in the top three. Psychomotor measures are strongest overall, which matters for realism: a knowledge test does not replace a dribbling test, it sits beside it. But the 0.81 figure means the cognitive side of the game is a scientifically valid target for prediction, not a marketing story.

Ivarsson et al. 2020: a more cautious estimate

Ivarsson and colleagues, a group including Arsenal FC staff, pooled studies of psychological factors and future performance. Their summary is that the prognostic value of psychological factors is small but statistically significant, with perceptual-cognitive factors among the strongest in the set.

The difference from Murr’s d = 0.81 is a real disagreement in the literature, and it comes mostly from what was pooled and how “future performance” was defined. It is the reason Insporta describes effect sizes as moderate when talking to clubs and parents, rather than quoting the largest number available. The validation article explains why the definition of success changes the answer.

Höner et al. 2021: the benchmark study

The largest prospective talent study in Europe: 13,869 boys aged 12 to 15, sampled across Germany within the national talent programme. Players were assessed on nine predictors, five objective motor tests and four subjective coach ratings, and success was defined as reaching youth academy level within three seasons. The success rate was 9%.

Three lessons carry over to any assessment, including a knowledge test.

A multidimensional model works better than any single measure. A logistic regression with several predictors explained more variance than any one of them. No score, however well built, should be read alone.

Coaches add information that tests miss. Subjective ratings improved prediction even after objective scores were included. The coach sees things the instrument does not capture. A test result belongs next to a coach’s judgment, not instead of it.

Sample size is statistical power. Reinhard, Mann and Höner (2025) had 110 players and say plainly that future studies should use larger samples. A study of 13,869 took a national federation, dozens of researchers and years. Any platform that records sport-specific test results at scale over time is collecting the kind of data the field lacks.

Höner et al. 2023: the reliability benchmark

The 360-degree video decision-making test used in the 2025 paper was validated earlier on an older group, U17 and U19 players, N = 48. The results:

Property	Finding
Split-half reliability	r = .78
Prognostic validity	Associated with adult playing level several years later
Player feedback	Positive

The reliability figure is the one Insporta takes as a target. A test with reliability at or above r = .78 is at the academic standard for a sport-specific cognitive instrument. Reaching it is a matter of design, not assertion: a ten-question test is a short form, so any longer self-assessment built from Insporta’s item bank needs enough questions, balanced difficulty across items, and a test-retest check on the same group a week or two apart before any reliability figure is quoted. Until that is done, Insporta reports what it has measured and no more, in line with its research principles.

Realistic expectations

Putting the four studies together:

• Sport-specific cognition predicts future level with a moderate to large effect (Murr d = 0.81; Ivarsson: small but significant).
• It is one predictor among several; psychomotor skills are stronger, and a combined model beats any single score (Höner 2021).
• Base rates are low: 9% of a national talent pool reached academy level in three seasons, so even a good predictor will flag many players who do not progress.
• A test earns the right to be used for these purposes by demonstrating reliability around r = .78 and prognostic validity on its own data (Höner 2023).

A knowledge test, then, is a legitimate part of the picture and never the whole of it. The three-level hierarchy explains why sport-specific content is the right place to look; this evidence explains how much to expect when you find it.