Training Load Variables in Elite Youth Soccer: Is a Data Reduction Approach Consistent Across Different Age Groups?

A key issue for practitioners is the ability to handle the increased quantity of data provided by wearable microtechnology and the identification of which metrics provide actionable insights in players training responses. The aim of this study was to investigate the ability of principal component analysis (PCA) to reduce the number of variables assessed in a player monitoring program and verify the consistency of variables retained across different age groups of an elite youth soccer academy.

A PCA was conducted to reduce the dimensionality of training and match data recorded by 145 players from Under 15 to Under 19 squads. The variables assessed included Global Positioning System metrics, heart rate measures, and players session rating of perceived exertion values (n = 82).

Seven principal components were extracted for each age group, describing 67.6% to 68.7% of variability. Inconsistencies were observed in the number of variables retained (range: 24-28) and their loadings between the different age groups. These differences in metrics retained and strength of their contributions indicate that PCA outcomes cannot be generalized across the different age groups.

General themes and constructs of load were observed across the 4 age groups, including measures of volume and intensity for both internal and external loads. The inconsistencies show that employing a PCA approach with a wide array of variables may not be practical for use in an applied environment, where the application of a conceptual framework can aid the selection process.