Yes, it's partly because of group delay issues that I'm going to the trouble of trying to match the filter as optimally as possible. Ultimately, I want to box up a couple of these filters as 50R in/out 'bench test' filters with SMA connectors. So I want to try and get the return loss and the group delay to look fairly good across the centre of the passband.
I had a play with RFSIM99 to see if it could plot group delay when my s2p model was matched with the captap network. It seems to be able to do this based on the S21 data from RFSIM99. However, I had to adjust the number of display points to be a sub multiple of the number of points in my s2p files. My s2p file spans 20kHz with 801 data points and this has an aperture of 25Hz per step.
I had to adjust the display/analysis points to get a sensible group delay plot in RFSIM99. See the plots of group delay below in RFSIM99 and Genesys. Genesys is set to 801 analysis points
Note that this is taken with the captap matching network on either side of the filter in both cases. This should agree quite well with the group delay measured with the VNA once I build the matching network for real. However, all of this depends on how well I extracted that first S2p model from the VNA. I did it in a hurry but it should be OK