I think the answer is precisely the realization that heliocentrism is a convenient framework for specifying the correct relationships. If one formulated their geocentrism-with-epicycles model to be precisely equivalent to heliocentrism in all physically observable details, just with Earth as the origin to their coordinate system, and acknowledged this equivalence, then that would be perfectly correct as well, albeit unduly cumbersomely framed. But the useful thing is that insight, that what happens is equivalent to the clean heliocentric model, whatever coordinates you might prefer to think of it in terms of.
Is it possible for a geocentric, epicyclic model to get the 3D positions of the solar system bodies approximately correct, as opposed to just their angular motion with respect to Earth?