If the network token value is high in comparison to the tokens it hosts, the network is over-valued and vice versa, so the theory goes

This is more than meaningless, it's obscure and bizzare and definitely adulterated, if this metric is determinant to how token value is being derived then I would say they'll be misplacing the yardstick to measure the basic and right things that determine a token adoption or value, I think the NVTV is just to really get people to look into this metric style. But I'm really wondering how they'll really come up with such theories.