AV-Comparatives: Real World Protection Test - April 2016

A turnaround implies McAfee used to be desirable at some point in recent history.

 

nice to see avast up there and sorry ESET has slipped in recent years...

 

I hate it when graphs dont start at 0 to make the difference between comparisons look massive, it starts at 80% what the hell.

 

There you go:

http://chart.av-comparatives.org/chart1.php?chart=chart2&year=2016&month=4&sort=1&zoom=0

 

Yes, that usually puts results into perspective. IDK why this isn't default selection.

 

None of the links posted are the default selection. Default looks like this:

http://chart.av-comparatives.org/chart1.php

 

Avira is consistently at the top, unlike AVG.

 

The "Zoom" tab has 4 options

and everyone can read the graph as he wants.

I can't see where the problem is.

 

Yes, that's what I'm talking about. IMO default (not links on this forum) should be 0-100%.

 

much better, should be the default. I guess they use 80 as base by default to try and hype up the differences. A bit like ATMO trying to hype up minor things like archive scanning.

 

It's just av-c doesn't like ppl to see that ALL av's are pretty much the same, detection wise

 

No, the first link in this post was generated by anon. You can generate your own links with your preferred view on that page.

 

Hope Eset and Sophos both pick up their game.

 

It seems that less than 100 % in a test is unacceptable now for some Wilders users

IMO Eset is doing fine in those tests and even better in real usage.

 

It also depends on what test one looks at. For instance https://threatcenter.crdf.fr/?Stats

 

Are they not continuing with their statistics on detection rates of AV products? If I'm not mistaken, that looks like it's from August 2015. The award image at the bottom of the page, which is "dynamically generated every night", shows ESET as #1 on 13 August 2015.

 

Symantec 0%? Trend Micro 0%? Certainly quite remarkable... I tend to believe in results from many different tests...

 

I'm not certain that any of the sample files are actually executed. It sounds like it's an on-demand scan using VirusTotal.

https://threatcenter.crdf.fr/?Stats

"The principle of this test is very simple and can be made every day a quick opinion on detections of various antivirus products. This test was created solely for informational purposes. We take the most recent samples of our databases and we subject them to various antivirus products. Once out of the sample, we scan (with VirusTotal Online scanner) directly with the latest update of antivirus products to show their effectiveness.

Thank you please note and understand that this test is carried out every day automatically via reports generated by VirusTotal. This test is conducted and published an informational purposes only, and we do not guarantee the accuracy of the data. The method used in this test is only based on the analysis proposed by VirusTotal engine."

 

@ Charyb, thanks, but it is far from being accurate...

 

I agree. Not to mention outdated.

Here is an explanation of the testing procedure from -> https://threatcenter.crdf.fr/?Faq

 

The product is actually not bad. The problems with McAfee have never been about it's protection levels (which have always been "acceptable"). It's the other issues such as performance, bloat and less-than-satisfactory customer service that have caused problems for McAfee.

I use the 2016 range on a few computers and while it is sometimes heavier than I would like, it does work well and is fairly simple to use for all.

 

Many antivirus nowdays are getting "100 %" in real world protection test and yet we have millions of users with up to date software getting infected each day

I would love to see what module has "protected" the machine; I have a suspicious that some antivirus developers learned to cheat those tests using web filter and cloud protection.

 

It's easy to be 100% when true zero-day samples aren't used.

 

It is possible to be 100% when variants of existing malware are used. For new generations of existing families or new families of zero days it is very hard to acquire 100%.

On the other hand it is very hard for testers to get their hands on zero day (maximum one day old discovery in the Wild) in any variant, so my guess is that no test agency or institute uses true zero day as you call it.

On the other-other hand new generations and new families are rare, so using prevalence as a factor zero days are a non issue also (the risk of trashing your car is higher than trashing your PC with a zero day).

What is a false zero day according to your definition (when there are true zero days logic predicts that there are also false zero days).

 

Agreed about the difficulty by reviewers in acquiring true zero day samples. That's why I prefer to code my own- not as popular, but to my thought more realistic.