Reply to thread

Message

<blockquote data-quote="Andy Ful" data-source="post: 904989" data-attributes="member: 32260">This is normal due to statistical errors. There are 3 new malware threats created per second. So, daily you have about 260 000 new samples. For 2 month test, we have 15 mln samples. Even if 99.9% of samples are polymorphic or similar, then you have 15 000 sufficiently different samples (it is an example, the real number is not known).There were only 372 samples in the test (~2.5% of samples from my example). Testing something on 2.5% of samples (preselected from 15 mln samples) gives you some statistical error, which is probably greater than the difference in detection between Bitdefender and Microsoft.<img src="data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7" class="smilie smilie--sprite smilie--sprite130" alt="(y)" title="Thumbs up&nbsp; &nbsp; (y)" loading="lazy" data-shortname="(y)" />If I correctly remember the AV-Comparatives document about the test reliability statistics, then sufficient reliability can be obtained by testing several thousands of malware samples.Edit.From the AV-Test report (<a href="https://www.av-test.org/en/news/facts-analyses-on-the-threat-scenario-the-av-test-security-report-2019-2020/" target="_blank">Facts &amp; Analyses on the Threat Scenario:&nbsp; The AV-TEST Security Report 2019/2020</a>), there are currently 4.3 malwares per second and about 78% targeted Windows systems.About 3500 sufficiently new ransomware files were found in the period of this AV-Comparatives test (exactly 5225 in the Q1 2020 according to Kaspersky).In the first half of 2019, SonicWall Real-Time Deep Memory Inspection (RTDMI) technology unveiled 74,360 ‘never-before-seen’ malware variants.</blockquote>

[QUOTE="Andy Ful, post: 904989, member: 32260"] This is normal due to statistical errors. There are 3 new malware threats created per second. So, daily you have about 260 000 new samples. For 2 month test, we have 15 mln samples. Even if 99.9% of samples are polymorphic or similar, then you have 15 000 sufficiently different samples (it is an example, the real number is not known). There were only 372 samples in the test (~2.5% of samples from my example). Testing something on 2.5% of samples (preselected from 15 mln samples) gives you some statistical error, which is probably greater than the difference in detection between Bitdefender and Microsoft.(y) If I correctly remember the AV-Comparatives document about the test reliability statistics, then sufficient reliability can be obtained by testing several thousands of malware samples. Edit. From the AV-Test report ([URL="https://www.av-test.org/en/news/facts-analyses-on-the-threat-scenario-the-av-test-security-report-2019-2020/"]Facts & Analyses on the Threat Scenario: The AV-TEST Security Report 2019/2020[/URL]), there are currently 4.3 malwares per second and about 78% targeted Windows systems. About 3500 sufficiently new ransomware files were found in the period of this AV-Comparatives test (exactly 5225 in the Q1 2020 according to Kaspersky). In the first half of 2019, SonicWall Real-Time Deep Memory Inspection (RTDMI) technology unveiled 74,360 ‘never-before-seen’ malware variants. [/QUOTE]

Verification