for years I have tried to rationalize Perry reflex- in your post are you suggesting gluing up deflex then reversing and glueing up reflex?
I can show you the math if you want, but there is nothing magic about it. All you are doing is preloading the glue up so that it puts a bit of tension on the belly at the expense of greater tension on the back. There is no free lunch, you are mostly trading stresses around. This works because most of our bow woods are much stronger in tension than compression, so the back can stand the higher stresses with no problem while reducing the stresses a bit on the belly where it needs that help.
This is only worthwhile for wood bows, and probably not even for really good compression woods. I was working with red oak, so it needed all the help it can get.
Because of the way I did the glue up it also gets a bit more work out of the core wood because there is some preloading there when the belly and core are glued initially. Normally the core contributes very little to energy storage and sees low stresses.
In the thread I linked to I show pics of what I did. Going from memory, I glued the belly and core lams into about 6" of deflex, then flipped that assembly over and pulled it into about 3" of reflex and glued on the back lam. The reason for the deflex first is to maximize the tension on the belly without ending up with a huge amount of final reflex that makes the bow harder to tiller and string, etc.
Mark