>Ptolemaic epicycles are a perfectly good predictive theory of planetary motion to within the margin of error of observation for several thousand years of human history.

Except for Mercury. The math for Mercury's orbit didn't add up until general relativity came along and frame dragging could be taken into account[0].

This doesn't contradict your point, because Pluto's orbit was a sign of something that observable physics couldn't explain, pointing towards a more complete explanation. As far as I'm aware, there's no "hole" in physics that can only be explained by the existence of antigravity, and models in which it exists don't have the same or better explanatory power as current models.

The fallacy here lies in looking back at the advance of science and assuming that because models were inaccurate before, they must always be arbitrarily inaccurate. Rather, the more we learn, the less likely it becomes that everything we know is entirely wrong, because any new models have to explain every observable facet of reality that the current one does as well as fill in gaps.

[0]https://www.forbes.com/sites/startswithabang/2015/11/25/how-...

Edit: I meant Mercury, not Pluto, and the other relativity. Oops.