There is not much mystery here. Transmission lines that are mismatched at all create impedance transformations at every odd multiple of a quarter wavelength. They become transparent, meaning the impedance at the input is exactly the impedance at the output, at even multiples of a half wave.
Calibration at the input allows you to see the same load your transmitter sees. Calibration at the output will show the actual impedance of the radiator.
Ferrite beads and baluns do not affect either measurement unless they introduce loss or an impedance transformation. Baluns especially are misused as matching devices because, when not properly deployed, they introduce loss. Loss reduces reflected power by a 2X factor and lead to grossly misleading VSWR readings at the input side of the coax.
The nanoVNA is a wonderful tool for both.
Warren Allgyer - WA8TOD
Warren Allgyer