Predicting Agility and Identifying Football Player Profiles: A Machine Learning Approach

Abstract

Purpose

This study examined whether static flexibility assessments predict agility performance in elite football players and investigated performance-based player profiling using machine learning.

Methods

A total of 157 elite male football players (aged 18–33 years) from 7 African nations (South Sudan, Liberia, Tanzania, Uganda, Rwanda, Cameroon, and Nigeria) completed the Thomas Test, 90/90 Hamstring Test, and Illinois Agility Test. Given violations of normality (Shapiro–Wilk W = 0.900, P < 0.001) and variance homogeneity (Levene’s F = 1.752, P = 0.044), non-parametric tests were used as the primary analyses, with parametric procedures retained as sensitivity checks. Machine learning models (Random Forest and Gradient Boosting) were trained under stratified 70/15/15 splits with repeated 5-fold cross-validation. K-Means clustering was applied for unsupervised player profiling; a five-cluster solution was retained and validated by silhouette analysis and the Davies–Bouldin index.

Results

No significant flexibility–agility association was observed (Mann–Whitney U = 2156, P = 0.847, r = 0.02; independent-samples t-test T = − 0.1033, P = 0.918). Random Forest (R² = 0.008, 95% CI: −0.023 to 0.039) and Gradient Boosting (R² = −0.054, 95% CI: −0.087 to − 0.021) both failed to predict agility outcomes. Test Number (56.8%), Age (23.1%), and Playing Position (11.9%) dominated model variance, while all flexibility variables combined contributed less than 10%. Unsupervised clustering identified five latent performance clusters (one a singleton treated as an outlier), while positional analysis showed faster forwards (median = 18.9 s), intermediate midfielders/defenders (median = 19.8 s), and slower goalkeepers (median = 21.1 s).

Conclusion

Static muscle tightness tests lack predictive validity for functional change-of-direction ability in elite football players. Future research should prioritise dynamic, sport-specific, neuromuscular, and cognitive performance measures in athlete profiling.

Keywords

Sports / Agility / Muscle tightness / Football player / Machine learning

References

[1] Krolo A, Gilic B, Foretic N, Pojskic H, Hammami R, Spasic M, et al.Agility testing in youth football (soccer) players; evaluating reliability, validity, and correlates of newly developed testing protocols. Int J Environ Res Public Health, 2020, 17(1): 294.

[2] Little T, Williams AG. Specificity of acceleration, maximum speed, and agility in professional soccer players. J Strength Cond Res, 2005, 19(1):76-8.

[3] Brughelli M, Cronin J, Levin G, Chaouachi A. Understanding change of direction ability in sport: a review of resistance training studies. Sports Med, 2008, 38: 1045-63.

[4] Križaj J. Relationship between agility, linear sprinting, and vertical jumping performance in Slovenian elite women football players. Hum Mov, 2020, 21(2): 78-84.

[5] Sheppard JM, Young WB. Agility literature review: classifications, training and testing. J Sports Sci, 2006, 24(9): 919-32.

[6] Paul DJ, Gabbett TJ, Nassis GP. Agility in team sports: testing, training and factors affecting performance. Sports Med, 2016, 46: 421-42.

[7] Behm DG, Blazevich AJ, Kay AD, McHugh M. Acute effects of muscle stretching on physical performance, range of motion, and injury incidence in healthy active individuals: a systematic review. Appl Physiol Nutr Metab, 2016, 41(1): 1-11.

[8] Maloney SJ, Richards J, Nixon DGD, Harvey LJ, Fletcher IM. Do stiffness and asymmetries predict change of direction performance?. J Sports Sci, 2017, 35(6): 547-56.

[9] Alexe DI, Čaušević D, Čović N, Rani B, Tohănean DI, Abazović E, et al.The relationship between functional movement quality and speed, agility, and jump performance in elite female youth football players. Sports, 2024, 12(8): 214.

[10] Robles-Palazón FJ, McMahon JJ, Ayala F, Comfort P. Things to keep in mind when selecting physical assessments in youth soccer: correlations between test performances, interlimb asymmetries, and effects of maturation. PLoS ONE, 2024, 19(6):e0305570.

[11] Flore Z, Welsch G, Bloch H. Return to play assessment after lateral ankle sprains – German male elite youth football (soccer) academy baseline data. Int J Sports Phys Ther, 2024, 19(8):976-88.

[12] Fox AS, Bonacci J, McLean SG, Spittle M, Saunders N. What is normal? Female lower limb kinematic profiles during athletic tasks used to examine anterior cruciate ligament injury risk: a systematic review. Sports Med, 2014, 44: 815-32.

[13] Aloui G, Hermassi S, Hayes LD, Bouhafs EG, Chelly MS, Schwesig R. Loaded plyometrics and short sprints with change-of-direction training enhance jumping, sprinting, agility, and balance performance of male soccer players. Appl Sci, 2021, 11(12): 5587.

[14] Kadlubowski B, Keiner M, Hartmann H, Wirth K, Frick U. The relationship between change of direction tests in elite youth soccer players. Sports, 2019, 7(5):111.

[15] Young W, Farrow D, Pyne D, McGregor W, Handke T. Validity and reliability of agility tests in junior Australian football players. J Strength Cond Res, 2011, 25(12):3399-403.

[16] Haugen T, Seiler S. Physical and physiological testing of soccer players: why, what and how should we measure?. Sportscience, 2015, 19: 10-26

[17] Claudino JG, Capanema DO, de Souza TV, Serrao JC, Pereira ACM, Nassis GP. Current approaches to the use of artificial intelligence for injury risk assessment and performance prediction in team sports: a systematic review. Sports Med Open, 2019, 5: 28.

[18] Jauhiainen S, Äyrämö S, Forsman H, Kauppi JP. Talent identification in soccer using a one-class support vector machine. Int J Comput Sci Sport, 2019, 18(3): 125-36.

[19] Rossi A, Pappalardo L, Cintia P, Iaia FM, Fernàndez J, Medina D. Effective injury forecasting in soccer with GPS training data and machine learning. PLoS ONE, 2018, 13(7): e0201264.

[20] Musat CL, Mereuta C, Nechita A, Tutunaru D, Voipan AE, Voipan D, et al.Diagnostic applications of AI in sports: a comprehensive review of injury risk prediction methods. Diagnostics, 2024, 14(22):2516.

[21] de Oliveira DCX, Gonçalves HR, Denardi R, Oliveira V, Deminice R. The influence of familiarization on performance testing of young soccer players. Motriz Rio Claro, 2024, 30: e10210188

[22] Paz GA, Maia M, de Sant’Ana F, Miranda H, Castro HJBP, Lima VP. Positional relationship between several performance tests and physical profile of Brazilian football athletes. Acta Sci Health Sci, 2019, 41(1):e43155.

[23] Díaz-Ochoa A, Gómez Renaud VM, Flores HJR, Hernández Cruz G. Variations in physical performance during a competitive season in Mexican female varsity soccer players by playing position. Retos, 2023, 49: 300-6.

[24] Savolainen EH, Ihalainen JK, Vänttinen T, Walker S. Changes in female football players’ in-season training load, intensity and physical performance: training progression matters more than accumulated load. Front Sports Act Living, 2024, 6: 1454519.

[25] Brown DJ, Arnold R, Standage M, Fletcher D. A longitudinal examination of thriving in sport performers. Psychol Sport Exerc, 2021, 55: 101934.

[26] Reilly T, Bowen T. Exertional costs of changes in directional modes of running. Percept Mot Skills, 1984, 58(1): 149-50.

[27] Zhang M, Shi H, Liu H, Zhou X. Biomechanical analysis of running in shoes with different heel-to-toe drops. Appl Sci, 2021, 11(24): 12144.

[28] Borg GBorg’s perceived exertion and pain scales, 1998, Champaign, IL: Human Kinetics

[29] Marcora S. Perception of effort during exercise is independent of afferent feedback from skeletal muscles, heart, and lungs. J Appl Physiol2009, 106(6):2060-2.

[30] Vanrenterghem J, Nedergaard NJ, Robinson MA, Drust B. Training load monitoring in team sports: a novel framework separating physiological and biomechanical load-adaptation pathways. Sports Med, 2017, 47: 2135-42.

[31] Lima-Alves A, Claudino JG, Boullosa D, Rangel Couto C, Teixeira-Coelho F, Pimenta ME. The relationship between internal and external loads as a tool to monitor physical fitness status of team sport athletes: a systematic review. Biol Sport, 2022, 39(3): 629-38.

[32] Bourdon PC, Cardinale M, Murray A, Gastin P, Kellmann M, Varley MC, et al.Monitoring athlete training loads: consensus statement. Int J Sports Physiol Perform, 2017, 12(s2): 161-70.

[33] Saw AE, Main LC, Gastin PB. Monitoring the athlete training response: subjective self-reported measures trump commonly used objective measures: a systematic review. Br J Sports Med, 2016, 50(5): 281-91.

[34] Buchheit M, Laursen PB. High-intensity interval training, solutions to the programming puzzle. Sports Med, 2013, 43: 313-38.

[35] Impellizzeri FM, Marcora SM, Coutts AJ. Internal and external training load: 15 years on. Int J Sports Physiol Perform, 2019, 14(2): 270-3.

[36] Carling C, Lacome M, McCall A, Dupont G, Le Gall F, Simpson B, et al.Monitoring of post-match fatigue in professional soccer: welcome to the real world. Sports Med, 2018, 48(12): 2695-702.

[37] Mendez-Villanueva A, Buchheit M, Simpson B, Bourdon PC. Match play intensity distribution in youth soccer. Int J Sports Med, 2013, 34(2): 101-10.

[38] Hulme A, Thompson J, Nielsen RO, Read GJ, Salmon PM. Towards a complex systems approach in sports injury research: simulating running-related injury development with agent-based modelling. Br J Sports Med, 2019, 53(9): 560-9.

[39] Kazemi M, Ingar A, Jaffery A. Injuries in elite Taekwondo Poomsae athletes. J Can Chiropr Assoc, 2016, 60(4): 330-41.

[40] Salazar-Martínez J, Valencia-Sánchez WG, Clemente FM. Comparative impact of small-sided games and high-intensity interval training on physical performance in youth soccer players. J Phys Educ Sport, 2023, 23(10): 2769-85

[41] Stanković M, Trajković N, Mačak D, Đorđević D, Lazić A, Milanović Z. Effects of linear and change of direction high-intensity interval training on physical performance of elite female soccer players. Biol Sport, 2024, 41(4): 31-9.

[42] Hermosilla-Palma F, Loro-Ferrer JF, Merino-Muñoz P, Gómez-Álvarez N, Zacca R, Cerda-Kohler H, et al.Optimizing muscle performance in young soccer players: exploring the impact of resisted sprint training and its relationship with distance covered. Sports, 2025, 13(1):26.

[43] Dong K, Jeong G, Chun B. The effects of different training interventions on soccer players’ sprints and changes of direction: a network meta-analysis of randomized controlled trials. Appl Sci, 2023, 13(1):446.

Rights & permissions

Beijing Sport University