But if he was making money on average (which he said he did), then bayesian logic would imply that the average outcome would be to make money. So it's not luck, it's the average outcome.

The "unlucky" and unusual event in this case is losing money on Bitcoin based on the prior evidence.