Investigating the coach's eye when evaluating and selecting 3 × 3 basketball players

Area of Science:

• Sports Science
• Psychology
• Data Science

Background:

• Coaches often rely on subjective data for player selection, despite recommendations for multidimensional approaches.
• Understanding decision-making theories is crucial for analyzing predictors in sports selection.

Purpose of the Study:

• To investigate talent aspects in 3x3 basketball player selection.
• To compare statistical approaches, specifically logistic regression and fast-and-frugal trees (FFTrees), for analyzing selection decisions.

Main Methods:

• Assessed 46 elite 3x3 basketball players using motor tests, anthropometrics, and coaches' subjective evaluations.
• Reduced variables and analyzed data with logistic regression and FFTrees.
• Examined interrater reliability using intraclass correlation coefficients (ICC).

Main Results:

• Subjective assessments were the most important predictor for selection in both logistic regression and FFTree analyses.
• Interrater reliability between coaches showed poor to moderate consistency (ICC: 0.38 for females, 0.51 for males).
• FFTrees visualized cue prioritization, reflecting bounded rationality in decision-making.

Conclusions:

• Player selection in 3x3 basketball is highly subjective, evidenced by low interrater reliability.
• FFTrees are suitable for modeling selection decisions under time and information constraints.
• Enhancing selection quality requires a team approach with shared criteria and actuarial support, like FFTrees.