Many data science competitions occur in the context of soccer match prediction. The Kaggle European Soccer (KES) database, one of the biggest soccer datasets available on Kaggle, includes information about soccer players and matches from season 2009 to 2015 in 10 different European countries. For what concerns players' performance indicators, sofifa experts' of Electronic Arts Sports are considered the leading authority: they state that specific abilities make up broader dimensions, each of which reflects a more general performance ability. In other words, players' performance attibutes (variables) of the KES database can be summarized into fewer performance composite indicators, useful for predictive modeling. Assuming experts' classifications solidity, Carpita et al. (Stat Model 19(1):74–101, 2019c) recently underlined the importance of variables transformation and information about players' role in building these indicators. However, previous works focused on clustering matches rather than players' attributes (e.g., investigating the role of seasonality in successful vs dropping performance; Wibowo in Commun Sci Technol 1(1), 2016), thus leaving the statistical examination of experts' groupings a still unexplored territory. The present work aims at shedding light on this aspect through the Cluster of variables around Latent Variables approach: this clustering method makes latent components simultaneously shine from variable groupings. This procedure might finetune the recently developed role-based players' performance indicators and improve predictive modeling of match outcomes. [ABSTRACT FROM AUTHOR]