No.
1. As you yourself have noted WAR is more directionally useful than precise. A 2 win difference is not in itself proof of anything. Also, defensive measures are less reliable than offensive ones - so WAR built off of defense is inherently "squishier"
2. I think valuable and "best" is enough. Otherwise, you are basically measuring the ability of a guy's teammates. Mike Trout does not deserve to be blamed for Arte Moreno wanting to pay Albert Pujols a kajillion dollars to be a traffic cone.
3. Where sabermetrics helps is to eliminate stupid mistakes - like Victor Martinez as a 2nd place MVP guy.
Personally you can't be MVP without creating a lot of "V". The top of the WAR list gives a solid list of nominees - something you can dig deeper on. For me that list is Trout, Betts, Lindor, Ramirez, Bregman, Chapman, and maybe Martinez. And THEN you can take in contextual factors, maybe some bonus for doing it in a pennant race, whatever. What is useful this year is Betts hits all the sweet spots - analytically, narrative-wise, best player on best team, 30/30.
The NL MVP is much much trickier. I'd give it to Javy Baez but that could go in any direction - including Jacob DeGrom though that won't happen.