Here are some photos of classics with roller cam followers resp. curved followers - which does not make any difference on cam shapes with same radius.
First is a Fifties Horex 400 standard cam, next my own designed and fabricated "sports" test cam with more lift, same acceleration, from computer calculated valve lift numbers from 40 curves in my bible. It had separate lobes for variation in spread, one lobe behind the indexing disc. I did the construction purely by graphical way, 10 times bigger drawing and follower plastic template plus degree disc, very laborious - you get the idea. No CAD anywhere near me 15 years ago.
I wonder if these Vincent roller arms have needle bearings inside, seem very small to me. If so, the outer race is very thin walled, would scare me a bit.
The prewar roller followers are Guzzi 1935, see Guzzi signed. And prewar Horex 600, modified for needle bearings by myself.
I do not see how there is a reverse rotation situation, a variation of rotation speed for sure. But so it is in the big end bearing due to the forward and backward swivelling of the conrod in one revolutiion at high speed ! That is why big end bearings have to be lightweight and I do not want rollers bigger than 4 mm in there. Alu cage or high tensile steel cage are acceptable, no brass as was used by some companies - that break after many decades from brass ageing.
Lastly still no advice from anyone on a slick way to design cam shapes starting from valve lift numbers and including all valve gear components ???
Vic