I want to evaluate a new antivirus product which obviously hasn't undergone the tests by known test labs. So I want to measure its performance in different aspects and come up with common comparisons and measurements that are available for popular AV products.
My question is that, are there any formal or standard methods, test environments, test data and information available to conduct such a test? Like getting a standard collection of malware samples and testing the new antivirus with that and comparing the results with available AVs or in an inferior case, comparing all possible antiviruses myself with available samples and environment.
Should I use virtual machines and sandox like technologies to see if the AV detects and clears malware samples and then revert back to a clean snapshot or are there easier ways to do it? Any kind of tip and guide will be appreciated.