Using Date Table Offsets Instead of Time Intelligence Functions

I've been tryign to limit my use of time intelligence functions and also calculate as I find they are generally better way of coding for various reasons.

One of the things I have found is that using offsets to do something like return sales for Last Year with the Financial Year Offset as - 1, is incompatible with a month granularity.

For example, we have the following measure.

Revenue LY NO CALCULATE FYear Offset =
VAR __sel_offset =

SELECTEDVALUE(Dates[FYearOffset])


VAR __sel_prev_offset =

__sel_offset -1


VAR __vTablePrev =

FILTER(
ALL(Dates),
Dates[FYearOffset] = __sel_prev_offset
)


VAR __LY_Sales =

SUMX(
__vTablePrev,
[Revenue]
)


RETURN

__LY_Sales

This works in a table with financial years but if I change the granularity of the table to month & year, the whole financial year amount is returned (example attached).

Because of this issue, I still find myself using DATEADD for LY calculations:

Revenue LY DATEADD =
VAR __LY =

CALCULATE(
[Revenue],
DATEADD(Dates[Date],-1,YEAR)
)

RETURN

__LY

As it appears to be the most flexible and easy to use.  However I realise the drawback in using TI functions as they are "black box" features that don't really let you see what is going on inside of them.

Does anyone know a way to solve for this while retaining the use of offsets?

I have worked out a way to do it using ISINSCOPE to see if we are on the month or year level in the table, but this method is quite verbose and appears unessesarily complex.