Observed pacing patterns and residual late-run variability in standardized hybrid fitness racing: A split-record framework for post-race performance monitoring in HYROX

HYROX offers a standardized split-record setting for examining within-race performance in hybrid fitness racing. This study evaluated observed pacing profiles and residual late-run variability as a descriptive post-race monitoring framework using public HYROX split records. The discovery sample comprised 61,541 athlete-event records from 89 events between 2021 and 2023, and a separately assembled secondary-source directional validation sample comprised 19,769 records from 237 events between 2021 and 2025. The primary E13_L68 model used Run 1–3 plus Work 1–3 as the early-race information window and Run 6–8 as late-run outcomes. Run 8 showed the clearest terminal slowing pattern. Record-level residual P90-P10 was 67.8 seconds in the discovery sample and 33.2 seconds in the secondary-source sample; final-run P90-P10 was 111.8 seconds and 63.2 seconds, respectively. Stricter 120-s and 150-s minimum run-split screens produced essentially unchanged estimates. Adding Run 4–5 and Work 4–5 reduced record-level residual P90-P10 to 47.6 seconds and 23.0 seconds. In strictly time-ordered Run 8 models that incorporated all available temporally prior Run 6–7 and Work 6–7 history, residual P90-P10 remained 67.2 seconds and 33.0 seconds, respectively. The secondary-source analyses showed the same directional attenuation with richer prior information. Descriptive final-run reference bands were stratified by division, sex, and early-ability tier. Residual late-run variability is therefore best interpreted as a model-dependent post-race performance indicator: richer prior race history explains a meaningful component of the variability but does not eliminate it. The construct should not be interpreted as longitudinal stability, a fatigue diagnosis, or a direct training-prescription threshold.