Yes for certain things like statistical trading (assuming some kind of "nice" Gaussian-like distribution) where you have lots of trades and just need to be more right than wrong it's probably useful.

Not here though, where you are trying to prove a (near) equivalence.