Guru: Angle of the limb beds will affect the pre-stress on the limbs, and it can change how much they're flexed when drawn. However, limbs could be mounted on the back of the riser coming straight off the bow as well.
Owl bait. They don't flex the same at the same draw length. All other things being equal, the tips aren't drawn back as far on belly mounted limbs. The farther limbs are drawn, the closer the limb tips become and the more acute the string angle, which contributes to finger pinch. Talking longbows and hyrids here, not recurves. Recurves can be drawn to nearly the breaking point before the tips begin to converge.
Here's what I think is the explanation for virtually identical performance/speed regardless of whether the limbs are mounted on the back or belly. Others who know more about this stuff can correct me if I'm wrong. For the sake of illustration, let's assume the difference between back and belly limb placement on the riser is two inches. It's not that much, but it's easier to work with round numbers. To achieve the same weight (say 50#@28 inches) as back mounted limbs, belly mounted limbs have to be 50#@26 inches because they're flexed two inches less at the same 28-inch draw length. This difference in actual limb weight probably wipes out almost all of the performance differences between back mounted and belly mounted limbs. The stiffer belly mounted limbs might actually give them a speed advantage. I don't know.
Another advantage of the belly mounted limbs, as others have already noted, is it's more difficult to torque the bow with a forward grip associated with belly mounted limbs. To the shooter, this is felt as a forgiving characteristic.
As I said before. Final differences aren't worth worrying about. Shoot what you like.