Difference in results of HFSS and ADS

Greeting Everyone!
I have designed and simulated a multilayer six port circuit made up of slot coupled quadrature couplers and dividers in HFSS 3D modeler (ANSYS Electronic Desktop). I used modal analysis and wave ports and achieved good and smooth magnitude and phase results in terms of S parameters.
I used 50 passes and 0.2 GHz step. Sweep type: fast
However, when i export the same multilayer structure to ADS layout layer by layer and then use port editor for excitation. In port editor, middle ground layer has been separately identified .
I get very ERATIC S-parameter results. I am using the same layer stackup as in HFSS and same rogers substrate RO4003C. Momentum simulation has been done.
0.2 GHz step has been set in both cases
Frequency of operation is 2.5-10 GHz.
Which results to trust? What am i doing wrong in ADS? Please find attached the snapshot of my HFSS and ADS results and port editor in ADS.
Thankyou very much

That looks ok.

We don't have enough information to see what your mistake was.

A popular mistake is to place a port on the edge of a wide ground polygon, so that the entire edge length becomes the port width. And that might then be too wide compared to the wave length. See https://muehlhaus.com/support/ads-ap...edge-area-pins