An improved methodology for estimating critical power from mean maximal power output data

Abstract

The aim of this study was to determine if Critical Power (CP) and W' can be estimated from mean maximal power output (MMP) data collected in cycling races. Data were collected from 13 under 23 professional cyclists (mean ± SD; age, 19.5 ± 1.1 y; body mass, 66.3 ± 5.0 kg; height, 180.0 ± 5.0 cm; CP, 5.7 ± 0.3 W · kg^-1). Participants conducted a CP test in the field to determine CP_Test and W'_Test. MMP data were then collected in races for the subsequent 90 days. CP and W' were estimated from MMP values in two ways, using fixed MMP durations, 2, 5 and 12 min (CP_Fixed and W'_Fixed), and via a novel filtering of second-by-second MMP data (CP_Filtered and W'_Filtered). CP_Fixed and CP_Filtered were not significantly different from CP_Test (Mean Difference (MD) 5 W and 7 W, respectively, p > 0.05). W'_Fixed and W'_Filtered were not significantly different from W'_Test (MD 2.68 kJ and 0.89 kJ, respectively, p > 0.05). CP_Fixed and CP_Filtered correlated significantly with CP_Test (r = 0.872 and 0.922, respectively, p < 0.0001 for both). Neither W'_Fixed nor W'_Filtered correlated significantly with W'_Test (p > 0.05). Both CP_Fixed and CP_Filtered provide valid estimates of CP_Test.; however, CP_Filtered provides a better estimate.

Keywords: Cycling; modelling; performance; power asymptote.