Subtract a SELECTEDVALUE row from a Matrix from all other Matrix Rows

[mrothschild]

PBIX file is here: https://drive.google.com/file/d/1YJ7I8PYqqEiap-mPjvprMlsLh9fv5Jkb/view?usp=sharing

In that file are three Matrices: 

• Cumulative Probability of Default - Selection
• Cumulative Probability of Default - All Ratings
• Cumulative Probability of Default - All Ratings minus Selection

The interaction  (via /Format/Edit Interactions) has been turned off for the Matrix, ". . ..All Ratings".  

My desired output is to subtract the figures in the ". . . Selection" from the figures in the ". . . All Ratings" so that the result is as follows:

Row Labels	12	24	36	48	60	72	84	96	108	120	132	144	156	168	180
AAA	(0.15%)	(0.34%)	(0.46%)	(0.69%)	(0.93%)	(1.17%)	(1.43%)	(1.65%)	(1.92%)	(2.18%)	(2.50%)	(2.74%)	(2.94%)	(2.98%)	(3.13%)
AA+	(0.15%)	(0.32%)	(0.54%)	(0.83%)	(1.12%)	(1.42%)	(1.69%)	(1.93%)	(2.20%)	(2.46%)	(2.74%)	(2.96%)	(3.13%)	(3.16%)	(3.31%)
AA	(0.13%)	(0.34%)	(0.51%)	(0.72%)	(0.92%)	(1.15%)	(1.35%)	(1.54%)	(1.78%)	(2.01%)	(2.27%)	(2.49%)	(2.62%)	(2.66%)	(2.81%)
AA-	(0.12%)	(0.29%)	(0.43%)	(0.70%)	(0.96%)	(1.21%)	(1.47%)	(1.72%)	(1.99%)	(2.26%)	(2.54%)	(2.76%)	(2.97%)	(3.02%)	(3.19%)
A+	(0.10%)	(0.28%)	(0.40%)	(0.62%)	(0.86%)	(1.12%)	(1.34%)	(1.53%)	(1.73%)	(1.92%)	(2.14%)	(2.28%)	(2.36%)	(2.28%)	(2.35%)
A	(0.10%)	(0.23%)	(0.38%)	(0.61%)	(0.83%)	(1.01%)	(1.16%)	(1.30%)	(1.45%)	(1.56%)	(1.74%)	(1.89%)	(2.00%)	(2.03%)	(2.08%)
A-	(0.09%)	(0.21%)	(0.34%)	(0.57%)	(0.76%)	(0.96%)	(1.07%)	(1.21%)	(1.41%)	(1.61%)	(1.85%)	(1.99%)	(2.10%)	(2.08%)	(2.19%)
BBB+	(0.06%)	(0.11%)	(0.12%)	(0.26%)	(0.37%)	(0.47%)	(0.59%)	(0.68%)	(0.74%)	(0.81%)	(0.92%)	(1.03%)	(1.08%)	(0.95%)	(0.91%)
BBB	0.00%	0.00%	0.00%	0.00%	0.00%	0.00%	0.00%	0.00%	0.00%	0.00%	0.00%	0.00%	0.00%	0.00%	0.00%
BBB-	0.09%	0.32%	0.68%	1.00%	1.36%	1.62%	1.84%	2.04%	2.13%	2.16%	2.21%	2.26%	2.33%	2.69%	2.83%
BB+	0.17%	0.60%	1.17%	1.62%	2.08%	2.52%	2.88%	3.08%	3.36%	3.64%	3.71%	3.93%	4.20%	4.45%	4.79%
BB	0.33%	1.15%	2.37%	3.41%	4.49%	5.26%	5.98%	6.57%	7.11%	7.55%	8.03%	8.37%	8.62%	8.86%	9.05%
BB-	0.81%	2.55%	4.42%	6.22%	7.76%	9.21%	10.40%	11.54%	12.36%	13.04%	13.46%	13.97%	14.49%	15.12%	15.59%
B+	1.83%	5.05%	8.23%	10.80%	12.75%	14.18%	15.49%	16.62%	17.61%	18.49%	19.19%	19.65%	20.20%	20.83%	21.32%
B	2.98%	6.98%	10.52%	13.26%	15.42%	17.35%	18.68%	19.63%	20.51%	21.38%	21.80%	22.29%	22.65%	23.07%	23.41%
B-	6.37%	13.32%	18.69%	22.23%	24.70%	26.45%	27.69%	28.62%	29.16%	29.57%	30.39%	30.83%	31.17%	31.64%	31.85%
CCC	28.15%	37.96%	42.83%	45.43%	47.31%	47.99%	48.81%	49.25%	49.60%	49.88%	49.99%	50.19%	50.51%	50.87%	50.73%

Screenshots of the above-referenced Matrices with the one at the bottom not working as intended:

I suspect I need to "clear" filter selections using ALL or ALLSELECTED, but I'm not familiar with the proper coding protocol for those.

THanks in advance!

[ELofstrom3C]

Not sure if this is what you're looking for but here's what I did:
1.) Create a calculated table (I called mine "Ratings") with the code "Ratings = DISTINCT('_Loss Given Default'[S&P Rating])".  This gives me a disconnected table with all the S&P values.
2.) Create a calculated measure in the new Ratings table with the code "Selected Rating = SELECTEDVALUE(Ratings[S&P Rating])".  This traps the value the user selects.
3.) Create a slicer with the Ratings[S&P Rating] field.  The trick here is that there are no relationships from the Ratings table to any other tables.

4.) Create a calculated measure in the Default table "Filter Me = IF(MAX('_Loss Given Default'[S&P Rating]) = [Selected Rating], 1, 0).
5.) In the first table filter on the "Filter Me" measure where the value is >= 1.

6.) In the bottom table filter on the "Filter Me" measure where the value is 0.

[mrothschild · Accepted solution]

@ELofstrom3C 

Thanks.  Your output looks like it resulted in the same, but started me down the path I needed.  Disconnected slicer was the key.  Here's the code that worked for my solution:

Probability of Default Delta = 

VAR _SelectedRows = 
    FILTER(
        ALL('_Loss Given Default'),
        '_Loss Given Default'[S&P Rating] = SELECTEDVALUE('S&P Ratings'[S&P Rating])
    )

VAR _ReferenceRatingProbDefault = 
    SUMX(
        ADDCOLUMNS(
            VALUES('_Loss Given Default'[Month]),
            "Probability of Default", CALCULATE(
                SUM('_Loss Given Default'[Probability of Default]),
                _SelectedRows,
                '_Loss Given Default'[Month] = EARLIER('_Loss Given Default'[Month])
            )
        ),
        [Probability of Default]
    )

VAR _UnreferencedProbDefault = 
    SUMX(
        GROUPBY('_Loss Given Default','_Loss Given Default'[Month],'_Loss Given Default'[Probability of Default]),
            '_Loss Given Default'[Probability of Default]
    )

VAR _Output = 
    _UnreferencedProbDefault - _ReferenceRatingProbDefault

RETURN 
    _Output