Abstract
Performance analysis in basketball often treats playing time as an aggregate of minutes, yet the same volume can be produced through different temporal architectures. We propose the stint, the continuous on-court interval between a substitution in and out, as the analytical primitive of competitive exposure: minutes quantify volume, stints reveal its architecture. Public EuroLeague play-by-play data from two regular seasons (2024 and 2025; 706 games; 65,657 stints; 15,307 player-games) were processed through a modular R workflow. A Structural Exposure Index (SEI) combined class-weighted non-linear duration, sequence penalty, and fourth-quarter amplification, alongside calibration-free structural ratios. Stints were classified as Micro (<1 min), Short (1-3 min), Moderate (3-7 min) or Long (≥7 min). Stint architecture was stable across seasons; Moderate stints dominated and 34.7-38.2% of SEI variance was orthogonal to minutes. Matched player-games differed by roughly one-quarter to one-third in relative SEI. Fourth-quarter share was negatively associated with next-game unadjusted Player Efficiency Rating (uPER) change, reaching significance in 2025 and pooled but not in 2024; maximum consecutive Long-stint run was associated with non-appearance in 2025 but not in 2024. The stint framework provides a replicable structural measurement layer, sharpening questions about rotation structure, role demands, and competitive exposure.