The fun thing about wheels is that there are likely dozens of pretty optimal solutions. When I was at Zipp still we were using an automated CFD approach developed by Matt Godo at Intelligent Light (now at STAR) to take an approach more akin to simulated annealing.. the idea being that there are likely multiple pretty highly optimized shapes that weren't really connected to each other very closely. Simulated annealing is nicely explained
HERE and visually described nicely by this gif and works to try and find various disconnected regions of possible solution.
The Reynolds wheels are interesting in that they have much further forward center of pressure than anything out there that I've seen, but that helps them offset the natural instability of that very sharp edge behaving as a leading edge, so the wheels require generally higher steering torque but have less torque variability with wind angle variability.
The ENVE solution is to have moderately linear torque with yaw angle, balancing the torque magnitude with torque variability and ensuring torque is always positive
The original solution here was the Firecest solution that I worked on, it was the first real handling model for a front wheel and the idea was that torque variability was fine as long as the magnitude of the overall torque was very low and all yaw angles had positive torque values..
Firecrest really changed the game as prior to that many shapes suffered from torque values that would oscillate between positive and negative depending on wind angle so a 3-5 degree change in effective yaw angle could change the torque on the wheel from positive to negative... so something like a passing car could cause dramatic swings in wheel torque that saw the rider having to reverse their reaction torque from one direction to the other very quickly.
http://www.SILCA.cc Check out my podcast, inside stories from more than 20 years of product and tech innovation from inside the Pro Peloton and Pro Triathlon worlds!
http://www.marginalgainspodcast.cc