TSQL Group-By Over Two Columns

Hi Thanks in advance for开发者_运维百科 helping. I have the following table:

create table temp_vol (symbol nchar(10), dseqkey int, volume float)

here are some values in the table:

symbol     dseqkey    volume (no commas in the actual values)
C          20672      422,764,512 
F          20672       47,144,600 
F          20671       51,695,000 
C          20671      371,950,496 
F          20670       48,680,600 
C          20670      654,208,512 
C          20669      626,121,600 
F          20669       93,489,504 
C          20668      712,674,304 

The output I am seeking is a 2-day simple moving average ((volume dseqkey + volume dseqkey-1) / 2) for each symbol, for each dseqkey. I can get the moving average working for all symbols across one dseqkey. I can get the moving average working for all dseqkeys across 1 symbol. I cannot seem to get both working together.

You could give each row a number using row_number(). That allows you to look up the previous row(s). Example edited for a 3-value moving average:

; with numbered as 
        (
        select  ROW_NUMBER() over (partition by symbol order by dseqkey) rn
        ,       *
        from    temp_vol
        )
select  cur.symbol
,       cur.dseqkey
,       avg(window.volume) as MovingAverage
from    numbered cur
join    numbered window
on      cur.symbol = window.symbol
        and window.rn between cur.rn - 2 and cur.rn
group by
        cur.symbol
,       cur.dseqkey
order by
        cur.symbol
,       cur.dseqkey

This prints:

symbol     dseqkey     MovingAverage
---------- ----------- ----------------------
C          20668       712674304
C          20669       669397952
C          20670       664334805,333333
C          20671       550760202,666667
C          20672       482974506,666667
F          20669       93489504
F          20670       71085052
F          20671       64621701,3333333
F          20672       49173400

Test data:

if OBJECT_ID('temp_vol') is not null
    drop table temp_vol
create table temp_vol (symbol nchar(10), dseqkey int, volume float)
insert temp_vol values
('C', 20672,  422764512 ),
('F', 20672,  47144600  ),
('F', 20671,  51695000  ),
('C', 20671,  371950496 ),
('F', 20670,  48680600  ),
('C', 20670,  654208512 ),
('C', 20669,  626121600 ),
('F', 20669,  93489504  ),
('C', 20668,  712674304 )

I'm not on a computer with a database engine, so I can't be sure about this:
UPDATED following comment

SELECT A.symbol, A.dseqkey, AVG(B.volume) MovingAverage
FROM temp_vol A
LEFT JOIN temp_vol B
ON A.symbol = B.symbol AND A.dseqkey BETWEEN B.dseqkey - 30 AND B.dseqkey
GROUP BY A.symbol, A.dseqkey