I'm not sure that the shift point should be at the top of the torque curve. The reason is that if your torque curve falls fast after reaching its peak, then rpms above the peak torque would having diminishing returns. I think shifting should occur when the torque in the next gear would be better than the torque in the present gear. Here is an example:
In second gear (and other gear for that matter), peak engine torque (490 ft-lbs) would be developed at 3500 rpms. Shifting from first to second will reduce engine speed by 40 percent, so the rpms in first gear would need to be 5800 rpms to have a second gear rpm of 3500.
Now the problem is that the torque in first gear at 5800 rpm is down to about 290 ft-lbs.
By waiting to shift out of first into second on the other side of the curve, i.e., the declining side, the engine will be producing better rpms and better torque values in the next gear. But I think you can wait too long.
Shifting out of first at 4000 rpms instead of 5800 rpms, torque in first gear would be 475 ft-lbs, second gear rpms would be 2400 rpms and torque in second would be 480 ft-lbs.
In summary, by waiting to shift so that the rpms in the next gear are at the peak of the torque curve, the present gear may be falling on its face.
To find out where to shift from first to second insofar as torque, one could graph out the torque curve for the two gears. The curves would be identical, except one be lagging on the horizontal scale by 40 percent. Where the two curves meet should be the optimum shift point.
Clear as mud

?
Now that I have confused even myself, that brings me back to horespower. Horsepower is really just a function of torque (or vise versa). In theory, shifting should be centered around torque and not hp. So why the obsession with hp and not torque?