Per page 40 in the 3.1.2 implementation guide, your observation is correct. Study Day 1 is the “Subject Reference Start Date”, so the day prior would be Study Day -1 (indicating one day before the DM RFSTDTC value). Study Day 0 does not exist, simply because it contains no semantic value given the assignment of Study Day 1 as the starting point.
I was a bit confused by this. Since *DY variables can definitely be negative but apparently not zero this would imply that the day before ‘Study Day 1’ is ‘Study day -1’ (i.e. study days are rounded away from zero). Is this the case?