‘Game Changers’: Discriminating Features within the Microstructure of Practice and Developmental Histories of Super-Elite Cricketers - a Pattern Recognition Approach

Electronic versions


  • Benjamin Jones

    Research areas

  • Talent Identification, Talent Development, Expertise Development, Microstructure of Practice, Skill Acquisition, Machine Learning, Pattern Recognition, Cricket, Rugby Union, Super-Elite, PhD, School of Sport, Health and Exercise Sciences


This thesis advances understanding of expertise development by addressing notable methodological issues, to become the first in field to quantitatively measure the influence of the microstructure of practice in the development of expertise in a sample of truly elite (super-elite) sportsmen, using machine learning techniques. The research protocol provides a means of bridging the existing gap between expertise development theory and research, and its application for talent identification and development (Baker, Schorer & Wattie, 2018; Holt et al., 2018). The thesis contains six chapters, including three research papers. Chapter 1 critically reviews the research on expertise development in sport to date and presents the rationale for the research programme, which aimed to overcome the theoretical and empirical limitations of this research, namely: (i) restricting investigation to comparisons of practice quantity; (ii) one-dimensional studies of individual expertise domains, disregarding the multifaceted nature of expertise; (iii) reliance on linear analysis techniques in identifying isolated precursors of expertise; (iv) assumptions of homogeneity within sports; and (v) inconsistent benchmarking measures for classifying expertise (Coutinho, Mesquita & Fonseca, 2016; Jones, Lawrence & Hardy, 2018; Schorer & Elferink-Gemser, 2013). Chapter 2 presents two studies to determine whether the relative age effect (RAE) observed in youth sport, extends into ‘super elite’ performers (Cobley, Baker, Wattie & Mckenna, 2009). The findings provide new evidence of RAEs at the super-elite level, presenting both inter and intra-sport differences (Jones et al., 2018). The research developed and applied a set of stringent criteria to benchmark super-elite expertise, and considered inter and intra-sport differences, by assessing RAE prevalence across the disciplines/positions of cricket and rugby union separately. Potential explanations for the findings are explored, owing to the survival and evolution of the fittest concepts, which suggest that RAE is a contributing factor in the efficient turnover of performers who do excel in sport. Chapter 3 applies non-linear machine learning (pattern recognition) analysis to a set of 93 developmental features (variables) obtained from a sample of sub-elite and elite cricket spin bowlers. The analysis produced a holistic predictive model consisting of 12 developmental features, from 93 measured, that discriminated between the elite and sub-elite groups, with very good accuracy (85%). The 12-feature model highlights elite spin bowlers’ greater quantity of domain-specific practice. The external validity of this new multidimensional non-linear model is also tested. Qualitative data obtained was subsequently analysed to achieve a deeper understanding of the discriminating features. A working group of England and Wales Cricket Board (ECB) pathway coaches and practitioners were invited to scrutinise the interpretation of findings, producing recommendations for the wider game. Chapter 4 examines the predictive power of the nature and microstructure of practice activity in a comparison of super-elite and elite cricket batsmen, domains of expertise development previously unexplored simultaneously. The findings identify psychologically challenging skill-based practice, relatively early in the development journey (age 16), as a catalyst for progression to super-elite expertise. The study modelled the development experiences of the super-elite and elite by adopting non-linear pattern recognition techniques, producing a holistic predictive model containing 18 features, from a possible 658, that discriminated between the super-elite and elite batsmen with excellent classification accuracy (96.3%). Evidence for the external validity of this model is presented. The impact of the PhD, measured by its overall contribution to the ECB’s talent pathway processes, is presented in Chapter 5. Chapter 6 contains a general discussion of the theoretical implications of the thesis’ discriminating findings, and commonalities identified across the levels of expertise. Finally, the combined theoretical and applied value of the research protocol is further evidenced by its cross-sport application to research programmes recently commissioned by UK Sport and The Rugby Football Union.


Original languageEnglish
Awarding Institution
  • Lewis Hardy (Supervisor)
  • Gavin Lawrence (Supervisor)
  • Raphael Brandon (External person) (Supervisor)
Thesis sponsors
  • England and Wales Cricket Board Ltd
Award date28 May 2019