Expose to the left?

[tho_mas]

There is at least one MFDB, which does display raw histogram (I forgot which one, but someone posted this on LL)

all Phase One Backs show the histogram with "linear response". There is only rudimentary software in the backs to show "something" on the display and this is the capture without tonal curves. There is an icc profile appiled that reproduces a neutral grey axis that is a TRC of gamma 1.8. But the histogram on the back and the RAW file in Capture One match exactly if you set the so called "film curve" in Capture One to "linear response" (in this case there is no correction of gradation or whatever).

[Panopeeper]

the histogram on the back and the RAW file in Capture One match exactly if you set the so called "film curve" in Capture One to "linear response"

This is a bad sign. I don't know C1, but I guess it does not display the raw histograms.

You can test this: shoot the very same scenery with vastly different WBs, like incandescent vs. fluorescent. If the displayed histograms change, then they are past WB application, thus not representative of the raw channels.

It is important to understand, that linear vs. color space curve histogram changes how far the histogram reaches at a certain level, but it does not change when it arrives at the right edge and indicates clipping. For example the linear histogram ends at the middle of the range if the exposure is 1 EV under raw clipping; the sRGB histogram is much farther to the right at the same exposure. Increasing that exposure by 0.5 EV expands the linear histogram to the 3/4 point, the sRGB histogram changes only a bit - but they both indicate clipping at the same exposure.

Both the WB application and the color space conversion affect the appearance of histograms a lot, no matter if linear or not, and both are rendering the histograms non-representative of the raw channels.

[tho_mas]

This is a bad sign. I don't know C1, but I guess it does not display the raw histograms.

you are right: it's not the RAW histogram.
However this is not "bad" - actually it's extremely consistant. The histogram on the camera shows exactly the histogram I see in C1 when no (so called) "film curve" is applied. As you always work with an input profile specifically for your camera in C1 it actually makes a lot of sense to display the profile corrected histogram on the camera. At least in the case of Phase One digibacks it makes sense... with DSLRs it's unavoidably different.

[Antimonite]

According to my experience clipping happens in RAW between +0.5 EV +1 EV respect RGB in the B or R channels, due to the colour space application and the WB correction. For this reason it is usually possible to overexpose if we want to improve the shadows while keeping the highlight information. We can ETTL if we have enough DR or ETTR if not.
In any case the RAW file process and the final image have to be HDR like, if we pretend to keep the original colours.