Weekly training loads are typically reported using absolute values and are not individualized according to competition positional demands (relative values). The aim of this study was to evaluate absolute and relative training loads and compare across playing positions during a full in-season in an elite soccer academy. 24 elite academy soccer players, categorized into five positions (CD: central defender [n = 4]; FB: full back [n = 5]; CM: central midfielder [n = 6]; WM: wide midfielder [n = 5]; FW: forward [n = 4]), were monitored using a global positioning system. Absolute training load was calculated using the total distance, the distance at moderate-speed ([15-20[ km·h^-1), high-speed ([20-25[ km·h^-1), sprint (> 25 km·h^-1), the total number of accelerations (> 3 m·s^-2) and decelerations (< -3 m·s^-2). Relative training load was calculated by dividing absolute training loads by mean values from the competitive matches. Training loads were determined daily according to their distance from match day (MD). One-way ANOVAs were performed to evaluate differences between playing positions. Absolute moderate-speed distance was greater for WM compared to CD (p = 0.015, and p = 0.017), while the opposite was shown for relative values (p = 0.014, and p < 0.001) on MD-4 and MD-3, respectively. The absolute moderate-speed distance was not different between CD, FB, CM, and FW, whereas relative values were greater for CD on MD+2 and MD-4 (p < 0.05). FB and WM performed greater absolute high-speed distance than CD on MD-4 and MD-3 (p < 0.05) while no difference was highlighted for relative values.Our results demonstrated that in the present academy players, training load for CD was underestimated using absolute training loads for moderate and high-speed distances. In contrast, relative training loads highlighted WM as an underloaded position. Therefore, relative training loads are recommended as they contextualize training loads according to competitive demands and favor training individualization.