You're right, I made the classic frequentist statistics mistake. All we can talk about with certainty are the probabilities of things happening under a specific null model.