Valve train geometry

YEP and lots of reasons why
BTW do not use the Lunati Method without thiking through what they are saying
How to Verify Valvetrain Geometry - Lunati Power
that's for ball stud rockers where you can raise and lower the fulcrum easily "The pattern should be centered on the valve tip" In a perfect world- get the spacers
, and as narrow as possible. got that right
If it is not, experiment with varying the pushrod length to yield the best pattern. uh what is the best pattern?
if a shaft rocker go as narrow as possible and ignore where the pattern is on the valve tip- unless it's way near the edge

Comp's method was worse last time I looked
If you aren't moving the fulcrum, you aren't correcting the geometry.

What determines how centered the sweep pattern is on the valve tip.

Exactly how does pushrod length affect the sweep pattern?

What determines how narrow a pattern is possible?

The Comp method is ignorance at its finest.