So does Copenhagen. That's why you can't choose between them.

Didn't you say you can't tell whether randomness is true or due to incomplete interpretation? If Copenhagen is incomplete, then it doesn't work, only its one part works - the Schrodinger equation.

MWI doesn't allow you to predict what you will end up observing, so it is just as incomplete.

It doesn't predict collapse if that's what you mean, because collapse doesn't happen, but it predicts that the result of observation is eigenvalue, and the prediction is consistent and matches observation.