Efficient early talent identification in soccer through safe deselection: Simulation and empirical evidence

Early-stage talent identification in soccer often involves evaluating large cohorts under severe time and resource constraints, increasing the risk of prematurely excluding players with high potential. This study reframes early talent identification as a problem of safe deselection , aiming to maximise screening efficiency while ensuring retention of high-performing individuals. We examined a common trial scenario in which coaches must reduce cohorts of ∼50 players within a single session. Using Monte Carlo simulations alongside empirical data from high-performance youth academies, we evaluated how different screening strategies influence the number of players that can be excluded without losing the top-performing individuals in small-sided games (SSGs). Simulated players were characterised by three correlated technical attributes (dribbling, passing, 1v1 performance), with SSG net goal difference as the performance outcome. Screening strategy had a large effect on efficiency (F = 13.49, p  < 0.001, ηp ²  = 0.42). Single-skill approaches allowed limited safe removals, whereas the three-skill composite enabled safe removal of up to 70% of players while retaining all top performers. Sensitivity across retention criteria showed consistent relative ordering of strategies. Empirical analyses confirmed that, when applying simulation-derived thresholds, all top 10% performers were retained at each stage of screening. These findings provide a proof-of-concept demonstration that combining multiple indicators can support efficient early-stage cohort reduction while minimising irreversible deselection errors.