No, evolution is based on testable predictions. For instance, if two species appear to be closely related (having a recent common ancestor), do their genomes when compared share more "spelling errors" (such as endogenous retroviruses) than do species that are less closely related?
Neverthless, evolutionary theory is ultimately based on untestable assumptions; eg., that mutations occur randomly, as opposed to being caused by an intelligent designer.
Similar testable hypothesis:
For instance, if two buildings appear closely related it is more likely that they have similar elevator configurations than two dissimilar buildings.
ML/NJ