Discerning excellence from mediocrity in swimming: New insights using Bayesian quantile regression

Area of Science:

• Sports Science
• Biomechanics
• Human Performance

Background:

• Previous research focused on average swimming speed predictors.
• The influence of somatic variables on swimming speed at different performance quantiles remains underexplored.

Purpose of the Study:

• To investigate if somatic predictors of swimming speed differ across quantiles (0.1, 0.5, 0.9) of performance.
• To identify key somatic variables associated with 100-m swimming speed across and within different strokes.

Main Methods:

• Utilized a Bayesian allometric quantile regression model with 363 competitive swimmers.
• Employed Bayes Factors and Leave-one-out cross-validation for model refinement.
• Analyzed somatic variables (arm-span, seated-height, shoulder-breadth, hip-width, calf girth) in relation to swimming speed.

Main Results:

• Arm-span, seated-height, and shoulder-breadth are strong predictors across strokes.
• Predictor importance varies by stroke and performance quantile.
• For elite swimmers (0.9 quantile), shoulder-breadth is key for front-crawl, hip-width for backstroke, and calf girth for butterfly.

Conclusions:

• Somatic variables are critical for talent identification in swimming.
• Matching swimmers to strokes based on somatic structure is important.
• Performance prediction models must consider the upper tails of distributions for elite talent identification.