The test would have been better if the undetected samples were ran to see how they were dealt with...
I don't thing that was what the test was about - it was a quick down & dirty direct comparison - how Avast and AVG detection rate compared, given that Avast is about to buy AVG.
Whilst running it would hopefully be compared against the avast cloud whitelist (hardened mode) and run against CyberCapture (CC). The problem is these files are on the hard drive and not being downloaded so may not be intercepted by CC.
This is where it becomes very messy, many things could depend on user settings as Hardened Mode isn't on by default. And who knows if AVG even has an equivalent.