It's awesome that you're working on this, it's definitely a hard problem to solve but one that would give great unique value to users. Feel free to bounce ideas with me, don't worry about being pedantic or anything I love it!

Yeah, how do you crowdsource prediction? How do you let users bet "this will win" while also allowing them to change their mind and bet the opposite? Especially, how do you do that with anonymous users (i.e. "sybil immunity").

I suppose I think about it in terms of the old puzzle with the two guards, one who always lies and one who always tells the truth. Your objective is to choose one of two doors (win/lose) and you get to ask one guard one question, not knowing which guard is lying. Then, the question you should ask is "What would the other guard say is the winning door?" and pick the opposite of whatever was responded.

So instead of somehow asking "Which door is the winner?", you ask "Which door do you think everyone else thinks is the winner?", in order to bypass liars entirely.