Hi   Is there a way to calculate 5 day rolling beta ie., returns relative to market return using Proc expand ? See data below - Ret is the individual stock return and mkt_ret is the market return   data temp; input DAT1 mmddyy11. ID $ RET MKT_RET ; cards; 9/1/2016 AA 0.0001 0.0001 9/2/2016 AA 0.000279 -0.002973 9/6/2016 AA -0.006091 0.000146 9/7/2016 AA 0.026914 0.002228 9/8/2016 AA 0.023175 0.025139 9/9/2016 AA -0.021908 -0.014465 9/12/2016 AA -0.023251 0.015054 9/13/2016 AA -0.034177 0.000588 9/14/2016 AA -0.032881 -0.010008 9/15/2016 AA 0.005656 0.003787 9/1/2016 ADM 0.0001 0.0001 9/2/2016 ADM 0.000023 -0.002973 9/6/2016 ADM 0.000094 0.000146 9/7/2016 ADM 0.000023 0.002228 9/8/2016 ADM 0.000023 0.025139 9/9/2016 ADM 0.000023 -0.014465 9/12/2016 ADM 0.00007 0.015054 9/13/2016 ADM 0.000023 0.000588 9/14/2016 ADM 0.000023 -0.010008 9/15/2016 ADM 0.000023 0.003787 ; run; If not, any do loop macros with proc reg would also be helpful.   thanks     ... View more

Hi    I'm trying to execute the below code to essentially compund 3 month return ..essentially rolling 3 month comp return for each ID but found a peculiar issue while calculating using lag3 function as you may note below. I've copied the output and also shown what the expected outcome should be for comp3_ret.  Could you kindly let me know where I'm missing out:    data pras2; input DAT1 mmddyy11. ID $ RET VAR1 VAR2 FLAG ; cards; 12/31/2010 A 0.20 25 100 1 11/30/2010 A 0.20 15 100 1 01/31/2011 A 0.20 35 100 1 08/31/2010 A 0.20 35 100 1 10/31/2010 A 0.20 20 100 1 09/30/2010 A 0.20 35 100 1 12/31/2010 AA 0.25 25 100 1 11/30/2010 AA 0.25 15 100 1 01/31/2011 AA 0.25 35 100 1 08/31/2010 AA 0.25 35 100 1 10/31/2010 AA 0.25 20 100 1 09/30/2010 AA 0.25 35 100 1 ; run; PROC SORT DATA = PRAS2; BY ID DAT1; data pras1; set pras2; by ID DAT1; retain comp; if first.id then comp = 1; comp = comp*(1+ret); /**** Trailing 3 month compounded return ****/ if (first.id eq 0) and (ID eq lag2(ID)) then comp3_ret = comp - 1 ; if (first.ID eq 0) and (ID eq lag3(ID)) then comp3_ret = (comp/lag3(comp))-1 ; run;   PROC PRINT DATA = PRAS1; format DAT1 MMDDYY10. ; RUN;   output : (see attached img)     ... View more