That still gives 50-50 odds with sufficient sample size, not much of a test of the research hypothesis (since many alternatives will predict the same direction)

That still gives 50-50 odds with sufficient sample size, not much of a test of the research hypothesis (since many alternatives will predict the same direction). It is better than 100% chance of rejection though.