I like this answer because of its circular logic (therefore impenetrable).

Simply declare a prior good state to be "the mean," then all we need to do is let mean reversion work its magic!