Does the Intent Match the Output: Aligning Development Goals With Training Load in Youth Basketball.

Multinomial logistic regression models classifying development goal category and specific goal type both achieved 66% classification accuracy.

• Both models achieved 66% classification accuracy with Kappa = 0.60
• Model for development goal category showed large reduction in AIC (ΔAIC = 1224.1)
• Model for specific goal type showed large reduction in AIC (ΔAIC = 2540.7)
• These large AIC reductions demonstrated that coach-assigned IDGs were associated with distinct external load profiles

High-intensity deceleration counts, vertical PlayerLoad, and high-speed running distances were key predictors retained in both classification models.

• Key predictors were identified following stepwise selection procedures
• The same predictors (high-intensity deceleration counts, vertical PlayerLoad, and high-speed running distances) were retained in both the development goal category model and the specific goal type model
• External load data were collected using Catapult Vector S7 devices during all on-court training sessions

Accumulated training time differed significantly across specific goal types across the season.

• Significant variation in accumulated training time was observed across the 16 specific goal types (e.g., cutting, shooting, load tolerance)
• This reflects systematic variation in emphasis across the season
• The 16 specific goal types were grouped under four overarching development goal categories: defensive, offensive, skill, and physical

Individual developmental goals (IDGs) assigned by coaches were retrospectively grouped into four overarching categories and 16 specific goal types using inductive thematic analysis.

• The four overarching development goal categories were: defensive, offensive, skill, and physical
• The 16 specific goal types included examples such as cutting, shooting, and load tolerance
• IDGs were assigned to each player at the start of each term by coaches
• Inductive thematic analysis was used to retrospectively classify the IDGs

External training load data were collected over two years from 25 elite male youth basketball players in a full-time residential academy.

• Sample consisted of 25 elite male youth basketball players
• Data collection spanned two years
• Players were part of a full-time residential academy setting
• Per-minute external load metrics were used in the regression models
• Catapult Vector S7 devices were used to record external load during all on-court training sessions