Joachim, you elucidated what I was thinking. The shape of a tip speed curve is important to what weight arrow it is attuned to.
It's unfortunate that we have a word for a rate decrease for speed (deceleration) but not one for a rate decrease for force (depression?!). A plotted out dynamic "depression" curve, along with the mass of an arrow would give you it's terminal velocity, and also the location in the release cycle where terminal velocity occurs. After that point the arrow is decelerating from friction. On a fast tipped bow with a reasonably massive arrow, that might be at or near brace height. On a very light arrow with a highly stacked bow with low brace tension, it might be well before that.
Since short limbs at high draw weight can have fast tip speed, I'm not surprised that fight shooting with light arrows works well with them. If a fast tipped bow also has low stack at the same draw weight, that would be ideal. A bow shape with short limbs might stack more than a similar bow with longer limbs, but the shorter limbs might have less inertia. Since two variables are present stack, and inertia, it's likely that two bow types will produce good results under different conditions (ie arrow mass). A longer version optimized for lower inertia where possible and extremely low stack, and a shorter version optimized for lower stack where possible and extremely low inertia. Then one or the other will be appropriate for a particular arrow type.
A very short bow with extremely low limb inertia, and some stack, shooting a very light arrow will depend on acceleration only in the early part of the release cycle. Because of this the results will be relatively inconsistent, and it will be highly dependent on the archer's release, bow condition, wind direction and resistance and many other factors.
A longer bow with extremely low stack, and some additional inertia, shooting a more moderate mass arrow will depend less on the early part of the release cycle. Because of this the results will be more consistent, the heavier arrow will gain more energy and moving inertia, be less affected by the above factors.
It may not shoot as far however as the best shot from the less stable alternative bow system. I don't know about actuals. In one case occasional records may be favored. In other cases consistent wins may be favored. It depends on whether you favor extremes or averages. Overall points, or one time maximums.
It seems to me that gentler recurves would likely reduce stack, and static and lever systems would increase initial force (accpression???!) early in the cycle.
It would be really useful to have a curve of limb tip speed. If that could be summed with the traditional F/D curve, a new curve could be generated which is much more characteristic of a bow. And if the arrow mass was known, the point at which terminal velocity could probably be determined, as well as total energy absorbed, etc.
These are just thoughts -- nothing proven. But just what seems reasonable to me.