You would expect about 1 per unflawed hash function tested. But the law of averages is less powerful when you focus on a specific hash. The chance of having no fixed points in a particular unflawed hash is 1/e, so there is probably a fixed point, and possibly more than one.