you mention low supply of tokens, but respect to what?
Respect to hardware objects and to coins, a digital crypto can be split in undefined number of decimals.

An ICO could issue 10000 digital coins as units, which could still be shared in 10⁸ bits customers.

Therefore, I suppose a good rule of thumb to judge about potential of economics value in ICO (which is different from financial trading) is the binding between a digital currency and a tangible output or service — which can be quantified and qualified as scarse or abundant, at least in order of magnitudes or by comparing similar things.

How do you qualify a supply of tokens as low?

Is it just perception or do you use a quantitative modelling to qualify a supply as low or high, (e.g. comparing supply of tokens between all existing ICOs, or more robust model) ?