The waveform is in principle quite correct, beacuse you have a phase reversal at each transition, but it seems that there is some amplitude issue. Just a couple of things:
1) the data-rate, in your simulation, is very high: the same as the carrier frequency. I think you should reduce it to have a better idea about amplitude distortion when the transition occours.
2) If you send to the modulator pseudo-random datas, instead of a clock, you should be able to see also the eye-diagram using accumulated plots.
Regards.