The VB100 test is a load of bull and only anyone who uses VB100 test to judge the reliability of a product doesnt have a clue about these stuff.
They use a very small sample of in the wild malware to test products. The sample is so small that it often becomes a matter of luck. No AV product provides 100% protection. Products can also fail if they provide false positives.
A better comparison is provided by av-comparatives.org and av-test.org
In those tests generally avira and kaspersky comes out on top (ignoring the ones like trustport and avk which uses multiple engines).