Hi,
I am just guessing at what your table looks like, burt it
sounds to me as if you store time period data and
measures (starting balance) in one table. Maybe creating
a separate table for time and another one as fact table
will work better.
Granularity in a cube does not change on the lowest level
as far as I know. So you should have one fact table per
cube.

Mixing time dimension and fact table is not a good idea
since you are bound to have several fact tables and have
to derrive the time dimension of each table separately.

If I totally misunderstood your point, please give an
example of your table.

THX Lutz