Predicting the Free-Weight Back Squat 1-Repetition Maximum: Comparing Regression Models and Minimum Velocity Thresholds Across Sexes

This study compared the accuracy of eight velocity-based one-repetition maximum (1RM) estimation methods during the free-weight back squat across sexes. Thirty resistance-trained individuals (fifteen males and fifteen females) completed two sessions involving four submaximal loads followed by actual 1RM attempts. The load-velocity relationship (LVR) was modeled using both multiple-point (40%, 55%, 70%, and 85% 1RM) and two-point (40% and 85% 1RM) methods. Four minimum velocity threshold (MVT) approaches were applied: (a) general MVT (0.30 m·s^-1), (b) actual MVT (the velocity at the actual 1RM in session one), (c) individual optimal MVT (the velocity that eliminated the difference between actual and predicted 1RM in session one), and (d) average optimal MVT (average of individual optimal MVTs across both sessions for each sex). Across sexes, only the individual and average optimal MVTs consistently avoided systematic overestimation of 1RM (raw errors ≤ 2.20 kg, p = 0.179-0.938) and produced relatively small absolute errors (4.5%-8.3%) compared with the other MVTs (8.0%-12.4%). Sex did not influence 1RM prediction accuracy (p = 0.636), whereas the LVR model demonstrated a borderline significant effect (p = 0.050). The optimal MVT did not differ between males (0.38 m·s^-1) and females (0.36 m·s^-1) but demonstrated poor reliability. Therefore, when direct 1RM testing is impractical, we recommend employing the unified value (0.37 m·s^-1) as an efficient alternative for approximate 1RM estimation across sexes. However, caution should be exercised when using velocity-based methods to estimate 1RM in the free-weight back squat, as all methods exhibited relatively wide limits of agreement.