An examination of signs, samples and subjective expert opinion as predictors of (de)selection in a youth male soccer academy in the UK

Barraclough, Sam ORCID: https://orcid.org/0000-0001-8584-6408, Till, Kevin ORCID: https://orcid.org/0000-0002-9686-0536, Kerr, Adam ORCID: https://orcid.org/0000-0003-4481-7496 and Emmonds, Stacey ORCID: https://orcid.org/0000-0002-2167-0113 (2025) An examination of signs, samples and subjective expert opinion as predictors of (de)selection in a youth male soccer academy in the UK. Journal of Sports Sciences. pp. 1-11. ISSN 0264-0414

Preview

Published Version Available under License Creative Commons Attribution Non-commercial No Derivatives. Download (3MB) | Preview

Abstract

Multidisciplinary profiling provides coaches with key information to augment their (de)selection decisions. These profiles often encompass objective and subjective data in the form of signs (isolated assessments), samples (contextualised assessments) and subjective expert opinion (SEO). Whilst multiple sources of information are considered by coaches during their decision-making, research exploring the extent to which objective and subjective multidisciplinary information can classify (de)selection is limited. Multidisciplinary data (physical profiling, match statistics, coach match ratings) were collected on 58 Under-16 (n = 20) and Under-18 (n = 38) youth male soccer players from a single academy in the United Kingdom. Group-level differences between selected (n = 39) and deselected (n = 24) players were explored, and binary logistic regression models were created to classify (de)selection. Analysis revealed a significant difference between selected and deselected players for match ratings (p < 0.0001), 505 left foot (p < 0.01), frequency of passes, percentage of successful aerial duels, and percentage of accurate crosses (p < 0.05). A classification model containing signs, samples and SEO data demonstrated the best model fit (AIC = 72.63), the highest discriminatory power (AUC = 0.79) and classified players with the greatest accuracy (78%) for (de)selection purposes. The use of signs, samples and SEO can support (de)selection decisions but fails to fully represent the complexity of the (de)selection process.