Sounds like an excellent target for an X prize. Say one that would reveal one set of data for submissions to match, and not reveal another similar set. The "theory/formulation/algorithm" with the best match for the second set, wins.

Obviously any kind of University contact would be valuable for your kind of quest; do you have any skills you can offer to help professors in the area? So that you could establish yourself a bit before trying to sell your idea? That requires a fair investment in time, I know, but I think that's what you're up against since so does evaluating theoretical ideas "just on spec."

I don't mean to contradict the previous comment, re familiarizing yourself with a journal's style, but that's a bit like teaching yourself Spanish without ever having anyone check your work. So there, too, an insider who can act as a first reader is invaluable.

I wish you luck, the professionalization of the academy is part of why innovation is now slowing IMHO.