DAX Query to MDX Cube

My Date dimension is based off a view in the database table.  One of the columns in the view is [Days from Today] which performs a DATEDIFF function to give each date in my dimension table an integer value.  Eg. yesterday = 1, the day before =2 etc.  This technique seems to provide a nice sliding date-range approach so users can simply refresh their reports/dashboards to see recent data without having to fiddle with filters.  I run a "ProcessUpdate" on my date dimension at the start of each day.

In Power BI, I like to set it to be "less than 31".  I capture the DAXQuery and run it using DAXStudio locally.  It seems to kill any data older than the filter setting.  I would have thought a Scoped measure in the MDX cube would still return a value but it won't

I've tried using LookupCube, LinkMember, StrToMember etc but it's like the session is scoping the cube down to a sub-cube so will never return a value.

Does this make sense?

The MDX Calculated measure is as follows:

CREATE MEMBER CURRENTCUBE.[Measures].[Purchases Previous Year]  AS NULL,

FORMAT_STRING = "#,###";  

SCOPE ([Purchases Previous Year]);  

    SCOPE([Dates].[Year - Week - Day].[Year]); 
        this = IIF([Purchases] <> 0,([Dates].[Year - Week - Day].currentmember.lag(1),[Purchases]),NULL); 
    END SCOPE; 

    SCOPE([Dates].[Year - Week - Day].[Week]); 
        this = IIF([Purchases] <> 0,([Dates].[Year - Week - Day].currentmember.lag(52),[Purchases]),NULL); 
    END SCOPE; 

    SCOPE([Dates].[Year - Week - Day].[Day]); 
        this = IIF([Purchases] <> 0,([Dates].[Year - Week - Day].currentmember.lag(52*7),[Purchases]),NULL); 
    END SCOPE; 

    format_string(this) = "#,###";   

END SCOPE;