Why is astrology not testable? Saying that the position of a certain celestial body or bodies at the moment of birth influences someone's personality is a hypothesis that can be tested, albeit a pretty vague one.
If we make it more specific by saying that (say) Piscians will share such and such personality traits as opposed to (say) Aquarians, then we can test it. We can just grab a bunch of randomly selected Piscians and Aquarians, run standard personality tests on them, and analyse the results statistically to see if there is an effect of astrological sign...
Also, if Aquarians are supposed to be in a particular good week for love or money or whatever, but Piscians are supposed to be having a bad time that week, we can just grab the same two groups, and use some dependent measure to see how their love/money/whatever situation changed (anything really, from asking them questions, to how their bank account balances change during that week, to standardized tests, to physiological measures). In order to make sure that the effect is the result of the position of stars and such during that week, run the study again when astrological predictions say that the Aquarians will have a bad week and the Piscians a great one, and you should see the reverse effect.
Of course, the researcher wouldn't be able to experimentally manipulate someone's sign or the position of bodies, but you could still test these hypothesis.
Not being a scientific field doesn't imply that that field doesn't generate hypothesis that can be disproved.