Looks to me like the formula for the helper columns is off by a factor of two. There would seem to be a couple, at least, of ways to rectify that situation, however in the following modification to that formula, I changed the constant -4 to -2:
Formula:
=IFERROR(INDEX(ROW($D$4:$D$75),LARGE(IF((($CT5=$F$4:$F$75)+($CT5=$I$4:$I$75)),ROW($D$4:$D$75)-ROW($D$4)-2,""),$AD$29-COLUMNS($D:D)+1)),"")
In the attached copy of the 'Sample' file the resulting values in DC5:DG24 seem to match those of the 'Rolling Averages' file.
Let me know if you have any questions.
Bookmarks