Using the hp peak as a shift point is a common fallacy. Having a graph of the hp curve is much more important. The idea is to select the shift point that has the same hp as the recovery rpm (rpm immediately after shifting).

For an example of how this works, suppose you have two engines that have a hp peak of 350 at 5200 rpm. On one, the hp after that point drops quickly, losing 50 hp in the next 500 rpm. On the other, the hp declines very slowly, staying within 10 hp for the next 1000 rpm, then dropping to 320 hp in the next 500 rpm. Should the shift points be at the same rpm for those two engines?

Next: the steps between the transmission ratios also affect the shift points. A wide ratio transmission requires a higher shift rpm than a close ratio transmission.

At the track, I use the dyno sheets and the transmission ratios to determine the best shift points, and use data acquisition to confirm that. The most recent adjustment was a drop in the driver selected 7200 rpm shift point down to a 6900 rpm shift point. It is almost impossible for a racecar driver to "feel" the right shift point as the torque will be dropping as the hp is still rising.