Several years ago I posted an experimental bow on here. My idea was to try and incorporate all the things into a bow that make it fast and then try to figure out how to best mitigate the issues that extreme designs create. I decided on a reflex deflex with huge rounded recurves with the tips about 12" behind the back of the bow. A design like this would normally be very unstable as it would have an unusually large amount of string contact with the limbs and very low string angles for most of the limb as well. My remedy for the lack of stability was to use a very small working limb section just outside the fades. That put a lot of strain on a very small area so I had to make it about 4" wide.
I knew the bow would break down fairy quickly but I just wanted to see if I could get a few shots off before it broke down. I only tillered it out to about 22" and extrapolated to weight to be about 50#. I hooked it up to my shooting machine with a scale inline and started shooting. The first shot right at 50# shooting a 500 grain arrow hit 212 and started dropping in speed and weight with each shot. After about 10 shits it had leveled off at about 38# and hitting 157 fps with a 500 grain arrow.
On the first shot it had stored about 125% of energy in relation to peak draw force. The last several inches of draw had a slight let off with no weight gain. Some of that was because of the geometry of the bow and some of it was most likely due to the bow taking set as I drew it further.