Confusion on out-of-gamut colors

I created a profile for an Agfa D-lab for Kodak Endura Supra. The target (Monaco Profiler 1728 patches) was measured twice with Eye-One Photo. The measurement data was brought into GMB MeasureTool and the average deltaE was 0.24 (Best 90%: 0.22 ; Worst 10%: 0.44).
So the measurements should be correct.

For benchmarking purposes, I downloaded a comparable profile from DryCreek and compared the 3-D (Lab) volume of both profiles in Chromix ColorThink. The 3D volume of the profile that I created was larger.

Next I opened a tricky image (with very saturated reds, yellows, blues) in Photoshop and although my profile has a larger volume it shows quite a lot of out-of-gamut colors (through gamut warning). The smaller volume DryCreek profile on the other hand shows practically no out-of-gamut colors?

How is this possible?

Andrew Rodney suggested the following to me on a forum:
“Do you have ColorThink? Id convert an image using both profiles and plot the gamut of the images. Use Adobe RGB (1998) and sRGB version of a test image. It could be the assumption built by the profile package as to the source for conversions.”

I was going to do so but unfortunately ColorThink Pro crashes when I try to open the image in ColorThink Pro (as I described in an earlier post).

Do you think too that it might be the “assumption built by the profile package as to the source for conversions” as stated by Andrew or are there other possible reasons?