After using the Coleman position, but using reverse as the basis, point it was better--however it was getting the back wheels wheels up and final tweaking that git it dialed in. The setting is much more finicky than I would have expected. Before pestering Coleman for their recommendation I had devised my own procedure that worked OK using the neutral gate s ground zero and adjusting the cable til forward and reverse were equally offset from neutral--, but using reverse seems to work better.
I always find it helps to know what you are doing--this is the transmission configuration (note the components are not shown in their normal alignment, this is a common annoyance in Asian power transmission diagrams):
The main (input) shaft is at the top--it is powered by the CVT driven sheave and drives the middle (counter) shaft in two ways. One is a chain drive (reverse)--'tother is a gear drive (forward). A sliding cogged shifting collar is what connects either the reverse driven chain sprocket or forward driven gear to the counter-shaft. It is the lateral position of that collar that we are adjusting.
Here is the entire forward/reverse sprocket/gear/collar assembly:
The collar is splined to the shaft,but free to move laterally. The reverse sprocket and forward gear are free to rotate on the shaft until engaged by the collar's cogs. Note that he cogs on the "forward" end of the collar are angled outward at their end, while the "reverse" end cogs are straight (or less angled). I guess this was done to ensure forward was a good solid engagement. However it seems to be what makes reverse engagement fussy.
The reverse driven sprocket rides on a roller bearing on the shaft, the forward gear does not. this is solid design as the reverse sprocket spends most of it's life rotating in the opposite direction of everything else. In "forward", the forward gear rotates with the shaft--no bearing is needed.
For completeness here's the reverse chain, a solid, robust so-called "silent" chain: