What's wrong with that? Epicycles are a good testable model. They were discarded when a better one was developed.
The problem was not that it was discarded, but what happened between new data and getting discarded and how long it took.
You don't want Epicycles because theories should be "simple" and Epicycles are added exceptions