I'm no expert on type systems so my definitions are not meant to be theoretically correct, just to describe the practical implications of static and dynamic typing.

What you're talking about seems to be more related to how type systems are implemented, i.e. that dynamically-typed languages are "unityped", such as the one and only "static type" of Python being the PyObject. Thus, when you "bind an unassigned variable and then assign a value later", you're actually implicitly declaring the type to be PyObject.

But it is deceptive to define type systems purely in terms of this implementation detail, because then it would appear as if dynamically-typed languages are untyped (i.e. "everything is one type" is the same as saying "there are no types"). Since we know languages like Python do some type-checking at run-time despite everything being a PyObject, it is more useful, I think, to think of these languages in terms of the practical result rather than in terms of the implementation.

But if that's not what you're saying, I apologize.