Variable table not containing rows from another table

That is true. Small update on my progress:

1) First I tried to extract from the FACT, all the subcategories that have a discounted product

VAR listofsubcat = SELECTCOLUMNS( FILTER( FACT, FACT[FLAG_Product_Discount] = 1), -- just a flag I did in the backend to identify easily the discounted products "DiscountedSubcat", RELATED(DIM[Subcategory Code]) )

2) Then I tried to calculate the SUM of prices for the remaining products, but whatever I am trying, I get a correct total sum, but in incorrect row level sum, by period:

I tried many variations to get the result similar to the below logic:

VAR listofsubcat2 = SELECTCOLUMNS(FILTER(DIM, DIM[Subcategory Code] IN listofsubcat), "_subcat", DIM[Subcategory Code])

RETURN SUMX( ADDCOLUMNS(FILTER(FACT, NOT (RELATED(DIM[Subcategory Code) IN listofsubcat2) ), "_price", FACT[Price] ), [_price] )

Here listofsubcat2 correctly extracts the subcategories which have a discounted product, if I hardcode the name in the measure in the NOT IN part, like that :

RETURN SUMX( ADDCOLUMNS(FILTER(FACT, NOT (RELATED(DIM[Subcategory Code) IN {"Subcat1", "Subcat2" etc) ), "_price", FACT[Price] ), [_price] )

--> it works, but of course I need the formula to be dynamic based on user's selection.

Tried also with calculatetable:

RETURN SUMX(CALCULATETABLE(FACT, NOT ( DIM[Subcategory Code] IN listofsubcat2 ) ), FACT[Price])

Same wrong result. Apparently Power BI is somehow overwriting my filter context while doing the individual row sum ? Also I checked further and even if some periods show the correct sum, if you do a countrows in Power BI vs the returned rows in SQL, the number is different so probably "behind the curtains" they are wrong as well.

Any idea what am I doing wrong and how can I make Power BI respect my filter without having to manually hardcore the values that I want to exclude ?

Thank you !