>NumPy (and Pandas) is still missing a proper missing value (NA).

But if it's missing a missing value, doesn't that mean that it has a proper missing value?

I'll let myself out now...