Excel SUMPRODUCT, INDEX function pairing

I am attempting to develop a actual vs. budget worksheet that will aggregate for the user the number of months activity based upon the month of the fiscal year entered. I believe this can be accomplished using a combination of the SUMPRODUCT and INDEX functions; however, I can't seem to get the context correct. I'm attaching a boiled-down version of the workbook here for your reference. The full worksheet contains over 1,000 account numbers and additional calculation (margins, rate of return, etc.) based upon the YTD actual and budget numbers returned. The worksheets in the file are:

CONTROL - where the user will input the fiscal year month (1 - 12) for the data desired. "1" represents the 1st month of the fiscal year, June, "2" represents the 2nd month of the fiscal year, July, etc.
BUDGET - Budget data
ACTUAL - Actual month's results
YTD VARIANCE - Worksheet that accumulates # of months budget and actual based upon users' input in CONTROL

The formulas I'm trying to fix are in YTD VARIANCE, columns C and D. They currently contain a SUMPRODUCT to look up a specific cell reference. I would like these to pull from the respective ACTUAL or BUDGET sheets based upon the user-entered value in CONTROL, cell B3 (e.g. 1=June, 2=July fiscal YTD, 12=May fiscal YTD, etc.).

Any suggestions how to make this work with a SUMPRODUCT/INDEX combination? Also, any suggestions for this same result using defined names would be appreciated.

Thank you,
John Countzler

No need for SUMPRODUCT here, you can use SUM(INDEX:INDEX) using MATCH within the INDEX to generate the correct ranges, eg:

Please Login or Register  to view this content.

The Budget will be as above but obviously referencing the other sheet.

On an aside if you can I would advise adding a column (hidden if preferred) that holds the MATCH results so you need only calculate once per row (or twice if format of Actual/Budget not mirrored exactly) - rather than calculating the MATCH twice per cell per row - obviously more efficient.

EDIT: Just to pre-empt, others may suggest using OFFSET which is also viable but it's Volatile so if your model is big I would advise against it myself...