Training at the optimum power load (OPL) is an effective way to improve neuromuscular abilities of highly trained athletes. The purpose of this study was to test the effects of training using the jump squat (JS) or Olympic push-press (OPP) exercises at the OPL during a short-term preseason on speed-power related abilities in high-level under-20 soccer players. The players were divided into two training groups: JS group (JSG) and OPP group (OPPG). Both groups undertook 12 power-oriented sessions, using solely JS or OPP exercises. Pre- and post-6 weeks of training, athletes performed squat jump (SJ), countermovement jump (CMJ), sprinting speed (5, 10, 20 and 30 m), change of direction (COD) and speed tests. To calculate the transfer effect coefficient (TEC) between JS and MPP OPP and the speed in 5, 10, 20, and 30 m, the ratio between the result gain (effect size [ES]) in the untrained exercise and result gain in the trained exercise was calculated. Magnitude based inference and ES were used to test the meaningful effects. The TEC between JS and VEL 5, 10, 20, and 30 m ranged from 0.77 to 1.29, while the only TEC which could be calculated between OPP and VEL 5 was rather low (0.2). In addition, the training effects of JS on jumping and speed related abilities were superior (ES ranging from small to large) to those caused by OPP (trivial ES). To conclude, the JS exercise is superior to the OPP for improving speed-power abilities in elite young soccer players.