I don't think that's a very good point. If it were, then quantum mechanics itself would make no predictions whatsoever. But it does: it makes probabilistic predictions. And when you have large numbers (the number of fluctuations in the early universe was very large), the statistics of those numbers becomes highly predictable, even if, in principle, anything can happen.
The primary issue that he does have a good point on is, to me, the measure problem. I am extremely
skeptical of his discussion about the cyclical universe, however, as that seems to completely violate everything we know about thermodynamics.