Linear skating speed key performance indicators in ice hockey: global or cohort-dependent?

The purpose of this investigation was to determine whether off-ice key performance indicators (KPIs) of linear skating speed are global across all skaters or modulated based on relevant cohort-dependent covariates. A total of 112 development- and university-level hockey players completed on-ice (30-m skate; ICE 0–30 m) and off-ice assessments (30-m sprint with split times, countermovement jump; CMJ, broad jump, and maximum chin ups). A linear regression model was created to predict ICE 0–30 m times from off-ice inputs with height, body mass, age level, and strength level included as covariates. Model parameters were estimated using the LASSO method with k-fold cross validation. The final model had a cross-validated R2 of 0.806. The strongest predictor of ICE 0–30 m times was 20–30 m sprint time (ß = 0.088). Relative propulsive mean power (ß = -0.064) from the CMJ, 0–30 m sprint time (ß = 0.058), and broad jump (ß = −0.046) represented second-tier predictors. Both relative braking net impulse (ß = 0.043) and relative braking mean power (ß = −0.009) from the CMJ were predictive factors for lower strength players only. The results indicate that top speed sprinting represents the primary global KPI and closest off-ice proxy for skating speed regardless of cohort.