Relative bench-press strength and jump power are associated with distinct tracking-based in-game shares in...

Ice hockey is an intermittent, high-intensity team sport, yet the transfer from preseason off-ice testing to tracking-derived in-game impact metrics remains underexplored in professional settings.

To examine associations between preseason off-ice fitness and tracking-derived 5-on-5 in-game impact in a Deutsche Eishockey Liga (DEL) team across three seasons.

Retrospective analysis of 28 male professional skaters (N = 50 player-seasons after listwise deletion, up to three observations per player). Predictors were 30 m sprint, countermovement jump (CMJ), relative single-leg strength (Bulgarian split squat 1RM per body mass, SLS/BM), relative upper-body strength (bench press 1RM per body mass, BP/BM), estimated maximal oxygen uptake (VO₂max), body fat percentage, and age. Outcomes were two team-relative share metrics derived from the league's official player-tracking system: the relative expected-goals-for percentage (xGF%Rel), defined as a player's on-ice share of expected goals minus the team's share with the player off the ice, capturing chance quality; and the relative Fenwick-for percentage (FF%Rel), the corresponding on-minus-off share of unblocked shot attempts, capturing territorial play. Ordinary least squares models included season fixed effects and player-clustered robust standard errors; multiplicity across predictors was controlled using the Holm procedure. Sensitivity models additionally adjusted for playing position.

Model fit was modest (adjusted R²: 0.16 for xGF%Rel; 0.21 for FF%Rel). After Holm correction, BP/BM was positively associated with xGF%Rel (β = .021, p_Holm = .011). For FF%Rel, CMJ was positively associated (β = .018, p_Holm = .012) and estimated VO₂max negatively associated (β = -.019, p_Holm = .035). Sensitivity analyses preserved the direction and approximate magnitude of all three associations. Statistical support was uniform only for CMJ-FF%Rel, which remained significant throughout; the BP/BM association did not survive allometric scaling and the estimated-VO₂max association fell just above the adjusted threshold after author exclusion.

In this single-team DEL cohort, preseason off-ice fitness showed metric-specific rather than global associations with 5-on-5 relative share outcomes. Relative upper-body strength aligned with chance-quality share, while lower-limb explosiveness aligned with territorial share; the inverse VO₂max association likely reflects role and deployment context rather than detrimental effects of aerobic capacity per se. Findings are associative, not causal, and warrant validation in larger multi-team datasets with explicit usage/context adjustment.