Issue with inaccurate results of measures

[bobbyd]

I have two issues in a simple report.     (I cannot use calculated columns, since I'm working off a shared dataset).

I have two measures created:

Calculated AGR =

SUM('Quarterly Reports'[L4_ELECTRONIC_CARD_SALES]) + SUM('Quarterly Reports'[L3_CARD_SALE_ENTRANCE_FEE]) +  SUM('Quarterly Reports'[L6_GROSS_RCPTS_PULLTAB_EVENT]) + SUM('Quarterly Reports'[L7_GROSS_RCPT_PULLTAB_INSTANT]) - SUM('Quarterly Reports'[L3_REGULAR_PRIZES_OVER_FIFTY]) - SUM('Quarterly Reports'[L4_REG_PRIZES_FIFTY_OR_LESS]) - SUM('Quarterly Reports'[L6_PRIZES_AWD_PULLTAB_EVENT]) - SUM('Quarterly Reports'[L7_PRIZES_AWD_PULLTAB_INSTANT])

AGR Match = IF(SUM('Quarterly Reports'[L8_TOTAL_NET_RECEIPTS]) = [Calculated AGR],"Match","NoMatch")

I'm wanting to filter results where AGR Match = "NoMatch"

Issue #1: The results of this show a few rows where the data is actually equal.  I've verified the underlying data and in fact the data is equal.  All columns are decimal with 2 decimal places.  The dataset is a Direct query dataset against an Oracle database, with the columns defined as number(12,2).  

Issue #2:

If I remove the AGR Match filter to show all rows,  the report sporadically shows rows which do not exist in the dataset.

For instance for quarter 20171, unit 1, there are no rows in the data.  Unit 1 exists but only for filing period 20051.   The blank values are incorrect and should not be displayed.

Removing the AGR Match column completely, then eliminates the incorrect rows from appearing.

My measures seem pretty simple, so what's wrong?

Bob