DOI: 10.62425/rses.1982533 ISSN: 2822-3527

Playing Time, Points, and Possession: Evaluating Individual Player Efficiency in the Turkish Basketball League through a Parallel Network SBM Data Envelopment Analysis

Adem Tüzemen, Çağdaş Yildiz
Show PDF Cite
This study applies a parallel non-homogeneous Network Slack-Based Measure Data Envelopment Analysis (NDEA-SBM) model to evaluate the individual efficiency of professional basketball players in the Turkish Basketball League (TBL) across three seasons — 2023–24, 2024–25, and 2025–26. Each player is modelled as a two-stage parallel system in which playing time is simultaneously allocated to offensive and defensive production. The offensive stage converts minutes, field goal attempts, and free throw attempts into points, made field goals, made free throws, and offensive rebounds. The defensive stage converts minutes into defensive rebounds, assists, steals, and blocks, with turnovers incorporated as an undesirable output. For player-season observations in which blocked shots per game equals zero, the corresponding slack term is excluded from the SBM denominator following Tone (2001) and Portela et al. (2004); this correction affects 68 of 300 observations in the defensive stage and prevents score distortion for positionally specialised players. Across 300 player-season observations, the pooled mean overall efficiency is 0.6519, and a persistent structural asymmetry is identified: offensive efficiency (0.7401) consistently exceeds defensive efficiency (0.5637) in every season examined, with the gap reaching 0.2081 in 2025–26. Slack analysis of the 261 inefficient observations identifies unused playing time and offensive rebounding shortfall as the primary sources of offensive inefficiency, while unused court time and defensive rebounding dominate the defensive stage. Super-efficiency analysis differentiates the 39 fully efficient observations. Panel analysis of the 16 players present across all three seasons reveals distinct efficiency trajectories that single-season statistics cannot capture.

More from our Archive