This is the idea I have for version 3 of my aerocap.
As far as I can determine within the limitations of Micro$oft Excel, there are no rearward flow angles that exceed 15 degrees, and most flow angles were designed to be around 12 degrees. That creasing you see in the second picture is an artifact of the 3D graph, and will not actually be a feature of the aerocap.
The idea behind this cap is to simultaneously transition the shape from the cab outline at the front of the cap to a semicircular outline at the rear, while dropping down from front to rear. Transitioning the shape appears to account for the vast majority of rearward angle changes.
I hope to see about a 13% or so improvement in FE over the baseline of no aerocap.
For comparison, here are versions 1 and 2.
Version 1 - destroyed - about 5% improvement in FE
Version 2 - currently in use - about 10% improvement in FE