Any single test with 400 samples per 2 months cannot be reliable. Anyway, the cumulative results of many such tests (two or three years of testing) can show the real differences between the real-world protection of some AVs. The differences between popular AVs are very small (probably negligible in practice).
Generally, these tests show that it is hard to be infected at home, except when one intentionally engages in risky actions (like searching for cracks, pirated software, etc.). The chances of infection are similar to the chances of severe injury in a traffic accident.
These tests do not show the direct detection of never-before-seen malware (especially payloads).