The uppercuts

The lead uppercut began as a script built from my photos. My spec: the load is a turn, not a squat. The hips and chest turn left over the lead leg, the fist drops to chest height, then the turn reverses from the feet up and the fist drives straight up into the chin. The targets were my idea too. Hitting a hanging bag from below makes no sense, so the uppercut got a pad facing down, then a heavy ball hanging at chin height. In the five punches work it hit 6 of 6 in the air, but light on the ball, about 10 N s. The speed work brought it to the target in 0.23 s instead of 0.36. Through the higher guard the ball stayed at 0 of 12, and the student distill4 still had the uppercut at one distance only, never aimed.

On 25 October I asked for both uppercuts over distance, and up, down, left and right. The lead one now blends the scripted uppercut at depth 0 out to my long left from my motion capture (take_02, 70.00–70.95 s) at +0.20 m. The rear uppercut was new: my short-to-long right from the same take (66.25–68.25 s, out to +0.31 m). Two teachers trained side by side on RunPod (work log). The rear one, upR1-209.7M, hit 54 of 54 in the air at about 10 m/s and 38 of 54 at the bag. The lead one, upL1, landed 40 of 54, never learnt the far end, and after the punch turned about 90° to its right and stayed there. The cause wasn’t my reference, which ends square at every depth. Training re-anchored “home” at every press and every recovery from a shove, and the opponent moved round to wherever the fighter faced, so nothing paid for facing one way. The fix keeps the heading for the whole episode, and the gate now measures the turn after every technique.

In the second round, upR2-209.7M ended square (3–17° three seconds after the punch) and was kept, at the cost of bag hits: 25 of 54. upL2 was square at the bag but still turned after the far uppercut into the air, and upL3, with a facing term more than three times as strong, was no better. Then I noticed upL2 hop on one leg after the far air uppercut. It didn’t fall, so the gate missed it, and every lead model did it, the student distill4 worst. upL4 added a penalty for a foot off the floor once the punch is over, and put half the presses in the air. The stagger went, and the far end was learnt at last: upL4-157.3M hits 54 of 54 in the air at 1.4–2.5 cm, with no steps after.

Both went into distill7 with the upright rear knee (work log). The kept student, distill7-245.8M, lands both in the air, 54 of 54 each, and the rear one 18 of 54 at the bag. But the lead uppercut ends 9–25° off square where its teacher ends 1–12° off. distill8 carries both on.

What’s open: every lead uppercut so far reaches the bag at about 4 m/s, under the hit speed, so it has no bag hits; and the student’s lead uppercut still turns more than its teacher’s.

What I think, as a coach

The reason for the distance work: “the head is not always exactly in the position that the animation is”.

On the long-uppercut video: “After the uppercut the fighter turns 90 degree to the right. It needs to keep facing the opponent.” On upL2, pointing at a screenshot of the far uppercut: “It lost balance here.” I chose the fix myself: pay for keeping the feet down once the punch is over, throw more of the far uppercut into the air, and keep the reach at 0.19 m.

On distill7’s videos, with both uppercuts in the student: “Looks good.”