Antenna simulation versus measurement in IE3D

I have designed a planar patch array for 5.8 GHz using IE3D. When I got the first manufactured antenna back, as one could expect it was not a perfect match, and I need to do a re-spin. What I want to do is to adjust the simulation parameters so that the simulated S11 matches the measured S11 of the antenna, and then verify that I have max antenna gain and min cross polarization in simulation at the same frequencies as the measurments. When this is accomplished I want to adjust the dimensions of the antenna for optimal performance at my wanted operating frequency.
To match the simulation to the measured I have first included a model for the SMA connector (made a coaxial tube, added a very small series resistor and shunt cap at the input). By doing this and adjusting slightly on epsilon_r and dielectric loss I can get a good match between simulated and measured around resonance (see attacment), but the simulated s11 curve is always longer than the measured. I.e. I can only match the curves at one frequency. Any advice on what parameters to adjust for abetter match?