Instead of answering your question directly, since I hear this concern about terminology so often, I'd like to offer how I think about this:
Theorem - a non-trivial observation that may or may not be easy to prove. That it's non-trivial means the proof is non-vacuous. In mathematical works, one finds theorems as provable claims of importance and that're taken in relation to the current line of reasoning, whatever it may be.
Lemma - a buttressing result proven in the course of attempting to prove a theorem. Eg., it may be necessary to prove claims A, B and C before being able to prove D, or say, it may be more helpful to prove the first three claims in order that the fourth be proven.
Corollary - a result that follows from a proven theorem. Usually it's a direct result, but sometimes such may require an additional discussion, in which case it may be indirect. Notwithstanding, in math, it's taken as a follow-along result.
Proposition - a claim that may or may not be related to a current line of reasoning and may or may not be proven. Typically, a proposition is found at either (a) the beginning of a line reasoning where it's used as a starting point or (b) out-of-context with respect to the current line of reasoning, which makes it part of a "hey, look at this!" type expression or (c) as an unproven axiom, in which case it might also be called a postulate.
I italicized the particular word proposition because I think it fulfills the 'Cicero criterion' of that which is most befitting; in other words, I think it applies to your situation. You don't seem to indicate that you're original goal was to author a scholastic paper, but rather that you stumbled on some finding in the course of your day-to-day work. For this reason, I'd assign it the label of proposition.
And now, assuming that your finding is non-trivial (ie., eg., it's not an identity proof), that there's an absence of evidence is no indication that there's an evidence of absence; indeed, another may have shown this to be false or true somewhere you've'nt come to experience. To that end, I'd suggest contacting an expert (or several) for verification because false proofs are very common. If your finding corresponds to something similar to polynomial uniqueness per spline interpolation, perhaps a numerical analyst, an optimization gal or an applied statistician could lend a hand. There's a host of specialties to appeal to here.
If indeed it's both non-trivial and provably true, then you might switch the labeling to theorem in the course of publishing your work, assuming that's what you'd do.