All input that goes into the devicechain gets halfed, even before the pre gain is processed. This headroom was added to avoid that you have to lower the master volume all the time when playing back more than one channel (the usual scenario when composing a song).
So its:
Sample&VSTiPlayback ->Sample&VstiGain -> *0.5 Magic -> PreGain (first device in the chain) -> DSP Chain -> Post gain.
Then maybe slider Amplify @ Sample Properties (and vsti) should be set by default at -6dB ???
I think any magic here is unecessary and rather misleading.