You'd have to sample many thousands of people who either have no symptoms or got sick. If the strain only exists in people with no symptoms then there's a good chance it's not deadly or at least less deadly than the other strains.

Of course you might find the best strain you can find exists in 100% of asymptomatics and 20% of the dead. That would indicate your strain has some non-zero death rate.

His assumption is based on live-attenuated vaccines. The measles vaccine being one of those. All the early vaccines where like that. The process only requires heat inactivation. So you end up with a dead or almost dead virus, which can't replicate anymore. Your immune system recognizes it as a foreign agent, makes antibodies to tag it for removal.

Today we prefer to immunize with a part of the virus structure, which end up doing the same thing.