Subtract values based on two slicer selections

[BIQuest321]

Hey All,

Been noodling this all day and coming up short so far. Essentially I will have two slicers that are supposed to offer dynamic comparisons of fact table data. These are basic calculations such as distinct counts. But I want to be able to select "Option A" from "Slicer A" and subtract "Option B" from "Slicer B". In the screenshot you see an example of two slicers, their distinct count below and the final piece I need is populating the math for the visual comparing the two. 

Very simple example of my fact table

ID	Name	Description
1	Aetna	A
2	Aetna	B
3	Aetna	C
4	BCBS Arizona	A
5	BCBS Arizona	B
6	BCBS Arizona	C

[bolfri · Accepted solution]

Create 2 tables:

Option A = GROUPBY(SampleData,SampleData[Name])

Option B = GROUPBY(SampleData,SampleData[Name])

Create 3 measures:

Option A Value = CALCULATE([Average ID],SampleData[Name] in VALUES('Option A'[Name]))

Option B Value = CALCULATE([Average ID],SampleData[Name] in VALUES('Option B'[Name]))

B minus A = [Option B Value] - [Option A Value]

You can add also 2 more bonus measures to prevent from selecting same name in A & B at the same time:

Not in A = IF(SELECTEDVALUE('Option B'[Name]) in VALUES('Option A'[Name]),0,1)

Not in B = IF(SELECTEDVALUE('Option A'[Name]) in VALUES('Option B'[Name]),0,1)

Put those measures on filters with condition = 1.

In this case you can't select Aetna, because it's on Option A, but if on option A will be Bolfri > Bolfri will dissapear from Option B, and Aetna will be avaliable to select.

PBIX file: https://we.tl/t-GVWssWxBpX

[BIQuest321]

Hey bolfri,

New day new problem. With the original premise basically still in tact here, how could I display the actual descriptions that are different or in common between A and B?

For example:
Option A has: A, B, C, D, E
Option B has: B, E, X, Y, Z

Result would return what they do or don't have in common:

Result (A,B): X, Y, Z
Result (B,A): A, C, D

[BIQuest321]

Is this possible without duplicating the fact table?? Obviously considering a much more complex scaled up example doubling the data is extremely performance prohibative. 

Edited - Okay I misunderstood your solution and my apologies on that. After reviewing your file this does exactly what I'm looking for without duplicating the fact table. Nicely done that's really slick!!