Just what does this CP parity really mean?

As far as I understand it, and I'm sure that's not far in comparison to many, CP parity means that not everything does what we expect when we expect it. So, those folks who build the BaBar machine and who like to look at these tiny little particles (or waves masquerading as particles . . . or whatever) have found this tiny kaon that doesn't fit in with their CPT (charge, parity, time" theorem. It's a symmetry theorem that says that all processes involving particles remain invariant if :
the particles are changed in antiparticles. So, the parts should act the same as the whole, I guess.
the system is reflected in a mirror
time is run backward
The funny thing is, those little kaons don't fit into that. They don't do the same thing if time is run forward as backward (since time currently doesn't go backward, I don't really get what they mean.) And not all the time, just some of the time. So I guess those folks at Stanford just spend their time looking at these kaons waiting for them to randomly do this thing that defies their theorem. Hmmmm . . .