Variability and predictability of performance times of elite cross-country skiers.

UNLABELLED Analyses of elite competitive performance provide useful information for research and practical applications. PURPOSE Here the authors analyze performance times of cross-country skiers at international competitions (World Cup, World Championship, and Olympics) in classical and free styles of women's and men's distance and sprint events, each with a total of 410-569 athletes competing in 1-44 races at 15-25 venues from seasons 2002 to 2011. METHODS A linear mixed model of race times for each event provided estimates of within-athlete race-to-race variability expressed as a coefficient of variation (CV) after adjustment for fixed or random effects of snow conditions, altitude, race length, and competition terrain. RESULTS Within-athlete variability was similar for men and women over various events for all athletes (CV of 1.5-1.8%) and for the annual top-10 athletes (1.1-1.4%). Observed effects of snow conditions and altitude on mean time were substantial (~2%) but mostly unclear, owing to large effects of terrain (CV of 4-10% in top-10 analyses). Predictability of performance was extremely high for all athletes (intraclass correlations of .90-.96) but only trivial to poor for top-10 athletes (men .00-.03, women .03-.35). CONCLUSION The race-to-race variability of top-ranked skiers is similar to that of other elite endurance athletes. Estimates of the smallest worthwhile performance enhancement (0.3× within-athlete variability) will help researchers and practitioners evaluate strategies affecting performance of elite skiers.