I happen to use Family Tree Maker, but I don't think my question is software specific.
I am using cluster genealogy to solve a particularly difficult problem. As a part of this process, I have defined a number of theories of varying levels of confidence that I am trying to prove or disprove.
I find it very useful to create a these theories or scenarios in Family Tree Maker because I sometimes need to see things put together to decide if it makes sense. However, I don't want to pollute my main family tree with guesswork. Currently, I've been creating additional alternate trees for each theory and sync'd them to ancestry.com. I've kept these trees in ancestry.com private so no one confuses them as being the Truth. The problem here is that there is a LOT of duplication between these various Theory trees and setting them up and propagating new facts across all of them is a pain.
What is the best way to go about modeling theories? Should I just do that outside the scope of my genealogy software?
The question on Transitioning from person-based genealogy to record-based genealogy? is related to my question, but not exactly. I know most of my people, I am not sure about their relationships.