As far as not having a source, maybe it is a bug/design problem with FTM in that it requires a SOURce record to create a source_citation. This is not a requirement in GEDCOM.
HOWEVER, even if you don't have a source for the ("ABT" plus date) in for example a birth, you can still use other sources that imply the information or lead you to the conclusion. For example:
Solution 1: You have birth sources for specific births of child1 and child3 and you have evidence that child2 was born between them. The accumulation of these three sources equal ("ABT" plus date) I would argue you should include all three sources under the birth fact for child2 then on the source_citation notes for one (or all three) that you used these 3 three sources to estimate the date.
Soultion 2: You could potentially create a source, similar to an interview, where you are the source of the calculation, or estimation. Then this, along with the other three sources are cited under the birth event/fact with the note for how you calculated the date on the source_citation connected to you as the source. You would only need to create the special SOURce once and reuse it every time you made calculation/estimation/determination that is not completely based on a tangible source.
Obviously these type of sources and citations are (or at least should not be) too common. But as we go farther back in time and/or rely on interviews, stories and hearsay as a portion of the data collected we need to still record the source of that information. But also make sure we indicate to the reader that it was a calculation, or estimation, or "crazy aunt Bess", because it could have some level of truth and be useful for later discovery.
This goes back to the primary problem I have with the concept of syncing a "work in progress" database from FTM to A.COM. I keep several databases, one for my "work in progress" and one for data that is, in essence, complete. Obviously, you are never truly complete, but if you are real close you can publish with an "as of" date.