Marc Garnica Caparrós

The collection of vast volumes of data in sports has allowed a strategic shift in performance analysis in basketball into a precise and data-driven era. Despite the exponential growth of advanced statistical analysis and predictive models to make informed decisions, these methods often rely on average tallies or simplified data models that may overlook the richer contextual and temporal information embedded in the raw data. This systematic review critically examines performance analysis approaches in basketball from the perspectives of data modeling and insights, assessing the alignment between the analytical techniques and the increasingly available fine-grained data. The main datasets available and the analytical methods proposed in 206 studies were extracted. While the data demonstrates an evolution in granularity, incorporating detailed event and tracking data, key findings reveal that box-score information and aggregated measurements remain predominantly used by practitioners. Only a limited number of studies have focused on exploiting the spatiotemporal characteristics of the raw datasets. Despite the efforts of various contributions to present a curated list of spatiotemporal analyses, the insufficiency of advanced analytical frameworks has constrained the effective utilization of temporal and sequential insights embedded in fine-grained datasets. Thereby, these techniques have not determined an influential shift in current research in basketball performance analysis, attributed to the reliance on data simplifications.