Why LMT over built-in color grading nodes?

I’ve been playing around with the Resolve ACES-friendly DCTL PFE (more acronyms please) and while it’s fun to use, I get better results simply using the nodes inside Resolve (curves, hue vs hue, hue vs sat, sat vs lum, etc) than trying to get CDL-type stuff to work. Typically I’m painstakingly recreating PFE’s using a bunch of curves in compound node groups all doing subtle things.

Is there any advantage to going through the trouble of writing my own little DCTLs for every operation I want to do rather than using built-in Resolve nodes? I get that I have explicit control over the implementation if I go through the pain of writing all my operations out, but are the built-in standard nodes in need of such a re-making in DCTL? Just trying to figure out what I’m missing by not going all-in on LMT’s.

I guess you could say I’m doing a lot of “empirical LMT” type stuff except that instead of analytical curve fitting I’m doing it by hand in the user-friendly curve editors I already have. Seems annoying to try to code it…

DCTLs are for very specific processes that you don’t have in the standard tools. No need to try to re-invent the standard tools when millions of colorists use those daily with great success.