I might be wrong, but I don't see that possibility. Although you can use Running Sum with a previous calculated value using LOD but not LOD on Something created with Running_Sum.

Please attach a workbook with dummy data

So that I understand you correctly - you want the distinct count of all customers that have made a purchase in the 10 months leading up to each month. January could have 1000 customers, February could have 850 customers, etc. where average customers per month would be something like 150.

Personally, I would approach this in SQL first. LOD calculations are great until you run in to table calculations. The reason for this is pretty simple - table calculations are computed as a last step in Tableau, where LOD calculations are completed first, and could not reference a table calculation that is computed later.

The LOD fails because it references your buckets which references a table calculation. If a table calculation is a dependency of any calculation nested under an LOD, it will fail.

Tableau executes:

1. The LOD calculation

2. The table calculation

2. The bucket calculation (conditional logic computed with the table calculation)

... and it has to work this way.

the table calculation being computed last can't be referenced by something computed higher up in order of operation.