rangestat question

[Curtis Chou]

Hellow
I have a question in rangestat
why do I always get the result that never calculates?
My mean is the result that is always the same with original

I want to compute MA5
rangestat (mean) v_debt v_roe v_Cdebt, interval(time -5 0) by(id time)

* Example generated by -dataex-. To install: ssc install dataex
clear
input float(time id v_debt v_roe v_Cdebt) double(v_debt_mean v_roe_mean v_Cdebt_mean)
19328 1          .           .           .                    .                   .                   .
19693 1  .18179545           .    .9389104   .18179544806480408                   .   .9389104247093201
20058 1   .6492594   -.7089842    2.163649    .6492593884468079  -.7089841961860657   2.163649082183838
20423 1  .04369804   -.7517949  .008138238  .043698038905858994  -.7517948746681213 .008138238452374935
20606 1   .2233313   1.0564739    .3494784    .2233312875032425  1.0564738512039185  .34947842359542847
20789 1 .028437067   2.5170796   .04255668  .028437066823244095  2.5170795917510986 .042556677013635635
20971 1   .2298042  -.21633975     .353778   .22980420291423798 -.21633975207805634  .35377803444862366
21154 1 -.04178981     .818469  -.28660178 -.041789811104536057   .8184689879417419  -.2866017818450928
21336 1   .3777859  -.24989977    .7478976   .37778589129447937 -.24989977478981018   .7478975653648377
21428 1   .4533736    .3821486    .4520029   .45337361097335815   .3821485936641693   .4520029127597809
21519 1   -.326355 -.024748646   -.4122111   -.3263550102710724 -.02474864572286606  -.4122111201286316
21609 1   .7289726   -.7508591     .622786    .7289725542068481  -.7508590817451477   .6227859854698181
21701 1   .9563724   1.2806365    .8218676    .9563724398612976  1.2806365489959717   .8218675851821899
21793 1   .2532618   .04117237    .4774139    .2532618045806885 .041172366589307785  .47741392254829407
21884 1  -.4537844    .2989276  -.53304607  -.45378437638282776   .2989276051521301  -.5330460667610169
21975 1 -.02242019   -.8486412 -.063347004 -.022420192137360573  -.8486412167549133 -.06334700435400009
22067 1 -.07704715    .7840909  -.22435866  -.07704714685678482   .7840909361839294  -.2243586629629135
22159 1  .07434082   .20955414   .29531816       .0743408203125  .20955413579940796   .2953181564807892
22250 1 -.23197924    -.325961  -.26288265  -.23197923600673676 -.32596102356910706 -.26288264989852905
22340 1 -.03853066   -.8265625  -.02780134 -.038530658930540085  -.8265625238418579 -.02780134230852127
end
format %dCY/N/D time

[Nick Cox]

You have daily dates all more than 5 days apart. So the means are just the original data. In fact doing calculations separately by(id time) enforces that too. by(id) is what you want, although it won't fix the first point.

[Curtis Chou]

the result still is the same
rangestat (mean) v_debt v_roe v_Cdebt, interval(time -5 0) by(id)

* Example generated by -dataex-. To install: ssc install dataex
clear
input float(time id v_debt v_roe v_Cdebt) double(v_debt_mean v_roe_mean v_Cdebt_mean)
19328 1          .           .           .                    .                   .                   .
19693 1  .18179545           .    .9389104   .18179544806480408                   .   .9389104247093201
20058 1   .6492594   -.7089842    2.163649    .6492593884468079  -.7089841961860657   2.163649082183838
20423 1  .04369804   -.7517949  .008138238  .043698038905858994  -.7517948746681213 .008138238452374935
20606 1   .2233313   1.0564739    .3494784    .2233312875032425  1.0564738512039185  .34947842359542847
20789 1 .028437067   2.5170796   .04255668  .028437066823244095  2.5170795917510986 .042556677013635635
20971 1   .2298042  -.21633975     .353778   .22980420291423798 -.21633975207805634  .35377803444862366
21154 1 -.04178981     .818469  -.28660178 -.041789811104536057   .8184689879417419  -.2866017818450928
21336 1   .3777859  -.24989977    .7478976   .37778589129447937 -.24989977478981018   .7478975653648377
21428 1   .4533736    .3821486    .4520029   .45337361097335815   .3821485936641693   .4520029127597809
21519 1   -.326355 -.024748646   -.4122111   -.3263550102710724 -.02474864572286606  -.4122111201286316
21609 1   .7289726   -.7508591     .622786    .7289725542068481  -.7508590817451477   .6227859854698181
21701 1   .9563724   1.2806365    .8218676    .9563724398612976  1.2806365489959717   .8218675851821899
21793 1   .2532618   .04117237    .4774139    .2532618045806885 .041172366589307785  .47741392254829407
21884 1  -.4537844    .2989276  -.53304607  -.45378437638282776   .2989276051521301  -.5330460667610169
21975 1 -.02242019   -.8486412 -.063347004 -.022420192137360573  -.8486412167549133 -.06334700435400009
22067 1 -.07704715    .7840909  -.22435866  -.07704714685678482   .7840909361839294  -.2243586629629135
22159 1  .07434082   .20955414   .29531816       .0743408203125  .20955413579940796   .2953181564807892
22250 1 -.23197924    -.325961  -.26288265  -.23197923600673676 -.32596102356910706 -.26288264989852905
22340 1 -.03853066   -.8265625  -.02780134 -.038530658930540085  -.8265625238418579 -.02780134230852127
end
format %dCY/N/D time

[Nick Cox]

Indeed; I said that such a change will not fix the first point.

You are asking for means across values on this and the previous 5 days. So for day 19693 that is the mean for daily dates 19688 to 19693. But there is only one observations in that window. And so on.

If you want the mean of this observation and the previous 5 observations, you need something different, such as

bysort id (time) : gen seq = _n

and then phrase your call to rangestat in terms of that variable.

If that's not what you want, I think you need to explain what you want.

(rangestat is from SSC, as you are asked to explain: FAQ Advice #12.)

[Curtis Chou]

Yes, I want the previous 5 observations
This is my original Code

cd "D:\操盤準備\財務data"
import delimited "All finance data.txt", encoding(Big5) clear
//ren 年月 time

egen id = group(簡稱)
destring 負債總額 流動負債 負債比率 流動比率 roea稅後 roe綜合損益, replace force
gen time = date(年月, "YM")

format time %dCY/N/D replace
order time

sort id time
by id: gen t = _n # I have use it

order time t


by id: gen v_debt = (負債總額/負債總額[_n-1])-1
by id: gen v_roe = (roea稅後/roea稅後[_n-1])-1
by id: gen v_Cdebt = (流動負債/流動負債[_n-1])-1


rangestat (mean) v_debt v_roe v_Cdebt, interval(t -5 0) by(id) # However, the result is the same

* Example generated by -dataex-. To install: ssc install dataex
clear
input float(time id v_debt v_roe v_Cdebt) double(v_debt_mean v_roe_mean v_Cdebt_mean)
19328 1          .           .           .                   .                     .                    .
19693 1  .18179545           .    .9389104  .18179544806480408                     .    .9389104247093201
20058 1   .6492594   -.7089842    2.163649  .41552741825580597    -.7089841961860657    1.551279753446579
20423 1  .04369804   -.7517949  .008138238   .2915842918058236    -.7303895354270935    1.036899248448511
20606 1   .2233313   1.0564739    .3494784  .27452104073017836    -.1347684065500895    .8650440422352403
20789 1 .028437067   2.5170796   .04255668   .2253042459487915     .5281935930252075    .7005465691909194
20971 1   .2298042  -.21633975     .353778   .2260542387763659    .37928692400455477    .6427518134005368
21154 1 -.04178981     .818469  -.28660178  .18879002891480923     .4524839346607526    .4384997789748013
21336 1   .3777859  -.24989977    .7478976  .14354444605608782     .5289980048934618   .20254119283830127
21428 1   .4533736    .3821486    .4520029  .21182370806733766     .7179885829488436   .27651863855620223
21519 1   -.326355 -.024748646   -.4122111  .12020932510495186     .5377848334610462    .1495703812688589
21609 1   .7289726   -.7508591     .622786  .23696523966888586  -.006871612121661504     .246275266011556
21701 1   .9563724   1.2806365    .8218676   .3580599458267291    .24262443805734316     .324290191133817
21793 1   .2532618   .04117237    .4774139   .4072352151075999     .1130750011652708   .45162614186604816
21884 1  -.4537844    .2989276  -.53304607   .2686401704947154     .2045462311555942    .2381355365117391
21975 1 -.02242019   -.8486412 -.063347004  .18934120330959558 -.0005854039142529169   .15224388365944228
22067 1 -.07704715    .7840909  -.22435866   .2308925138786435      .134221193070213   .18355262652039528
22159 1  .07434082   .20955414   .29531816  .12178722489625216     .2942900626609723    .1289746550222238
22250 1 -.23197924    -.325961  -.26288265 -.07627138774842024   .026523800566792488 -.051817050824562706
22340 1 -.03853066   -.8265625  -.02780134 -.12490346500029166   -.11809868117173512  -.13601959496736526
end
format %dCY/N/D time

------------------ copy up to and including the previous line ------------------

Listed 20 out of 39449 observations

[Nick Cox]

OK, so the code in #4 applles, Use seq not time in your call to rangestat.

 
 bysort id (time) : gen seq = _n 
rangestat (mean) v_debt v_roe v_Cdebt, interval(seq -5 0) by(id)

[Curtis Chou]

thanks very much

[Curtis Chou]

Thanks, it really different

* Example generated by -dataex-. To install: ssc install dataex
clear
input float(time id v_debt v_roe v_Cdebt) double(v_debt_mean v_roe_mean v_Cdebt_mean)
19328 1          .           .           .                   .                     .                    .
19693 1  .18179545           .    .9389104  .18179544806480408                     .    .9389104247093201
20058 1   .6492594   -.7089842    2.163649  .41552741825580597    -.7089841961860657    1.551279753446579
20423 1  .04369804   -.7517949  .008138238   .2915842918058236    -.7303895354270935    1.036899248448511
20606 1   .2233313   1.0564739    .3494784  .27452104073017836    -.1347684065500895    .8650440422352403
20789 1 .028437067   2.5170796   .04255668   .2253042459487915     .5281935930252075    .7005465691909194
20971 1   .2298042  -.21633975     .353778   .2260542387763659    .37928692400455477    .6427518134005368
21154 1 -.04178981     .818469  -.28660178  .18879002891480923     .4524839346607526    .4384997789748013
21336 1   .3777859  -.24989977    .7478976  .14354444605608782     .5289980048934618   .20254119283830127
21428 1   .4533736    .3821486    .4520029  .21182370806733766     .7179885829488436   .27651863855620223
21519 1   -.326355 -.024748646   -.4122111  .12020932510495186     .5377848334610462    .1495703812688589
21609 1   .7289726   -.7508591     .622786  .23696523966888586  -.006871612121661504     .246275266011556
21701 1   .9563724   1.2806365    .8218676   .3580599458267291    .24262443805734316     .324290191133817
21793 1   .2532618   .04117237    .4774139   .4072352151075999     .1130750011652708   .45162614186604816
21884 1  -.4537844    .2989276  -.53304607   .2686401704947154     .2045462311555942    .2381355365117391
21975 1 -.02242019   -.8486412 -.063347004  .18934120330959558 -.0005854039142529169   .15224388365944228
22067 1 -.07704715    .7840909  -.22435866   .2308925138786435      .134221193070213   .18355262652039528
22159 1  .07434082   .20955414   .29531816  .12178722489625216     .2942900626609723    .1289746550222238
22250 1 -.23197924    -.325961  -.26288265 -.07627138774842024   .026523800566792488 -.051817050824562706
22340 1 -.03853066   -.8265625  -.02780134 -.12490346500029166   -.11809868117173512  -.13601959496736526
end
format %dCY/N/D time