12:05 a.m.
Hi,I am trying to write code for the rolling windows forecasting using HAR-RV model.My
codes are:
set verbose off loop i=1..100 -q smpl 23+i 335+i ols RV5_FTSE const RV5_FTSE(-1)
HARWEEK HARMONTH fcast 337 437 1 forecasted --rolling endloop
I am not sure whether my forecasted values are rolling windows forecasts or not. I can see
on the script I succeeded re-estimated rolling windows regressions (that s why i did not
put -q there). But, I am not sure my one-step-ahead forecasted values are generated with
sliding re-estimated regressions. [I am not sure should I change my 5th line with fcast
336+i 336+i 1 forecasted --rolling] Or what should I do to get one-step-ahead rolling
windows forecasts in this loop command?
Does anyone know about that. I think I am missing something here on codes. I am sorry I am
making busy this line about this a couple of times, but I am trying to understand and
there is no one i can ask about that except here.
Thank you very much.
9:56 p.m.
Am 14.05.2020 um 02:05 schrieb Burak Korkusuz:
I am trying to write code for the rolling windows forecasting using
HAR-RV model.
My codes are:
Again I would ask you to send codes that can actually be run. You're
using undocumented and/or unavailable datasets in your examples. This
makes it difficult to help you.
set verbose off
loop i=1..100 -q
smpl 23+i 335+i
ols RV5_FTSE const RV5_FTSE(-1) HARWEEK HARMONTH
fcast 337 437 1 forecasted --rolling
Next, please stop using the option "--rolling". This is still working
for compatibility reasons but was renamed to "--recursive" in gretl
2017d (see http://gretl.sourceforge.net/Backward.html). This is
important because "rolling" was an ambiguous term. How do you define
rolling in your case, what exactly do you want to achieve?
cheers
sven
4:59 a.m.
Sorry, my fault, thanks for responding. I am adding my data to the email. I have 5-minute
realised variance data for FTSE.
Basically, I am trying to do rolling windows forecasting using the HAR-RV model (in time
series). I will have the initial sample; observation 24-336 (I labelled observation date
as numbers). Observations between 337-437 will be the out-of-sample. First, I want to
estimate HAR model with 24-336 obs. and then forecast one-step-ahead forecast [337]. Next,
estimate HAR with 25-337 obs. and then forecast [338]. Estimate with 26-338 and then
forecast [339] ...
I am not sure I am doing right with my codes and in my case rolling windows forecasting
is; one-step-ahead forecasted values are generated by each sliding different HAR model
(i.e. 24-336, 25-337, 26-338 ...). I am doing that with HAR-RV model and later on, I will
add to the model as an exogenous variable –VIX, EPU, Industrial production etc.– and
compare forecasting performance.
HAR-RV model is:FTSE.RV=CONSTANT+FTSE.RV(-1)+FTSE.WEEKLY.AVERAGE+FTSE.MONTHLY.AVERAGE
FTSE.WEEKLY.AVERAGE=FTSE[(-1)+(-2)...+(-5)]/5
FTSE.MONTHLY.AVERAGE=FTSE[(-1)+(-2)...+(-22)]/22
Thank you very much, On Friday, May 15, 2020, 10:56:40 PM GMT+1, Sven Schreiber
<svetosch(a)gmx.net> wrote:
Am 14.05.2020 um 02:05 schrieb Burak Korkusuz:
I am trying to write code for the rolling windows forecasting using
HAR-RV model.
My codes are:
Again I would ask you to send codes that can actually be run. You're
using undocumented and/or unavailable datasets in your examples. This
makes it difficult to help you.
set verbose off
loop i=1..100 -q
smpl 23+i 335+i
ols RV5_FTSE const RV5_FTSE(-1) HARWEEK HARMONTH
fcast 337 437 1 forecasted --rolling
Next, please stop using the option "--rolling". This is still working
for compatibility reasons but was renamed to "--recursive" in gretl
2017d (see http://gretl.sourceforge.net/Backward.html). This is
important because "rolling" was an ambiguous term. How do you define
rolling in your case, what exactly do you want to achieve?
cheers
sven
_______________________________________________
Gretl-users mailing list -- gretl-users(a)gretlml.univpm.it
To unsubscribe send an email to gretl-users-leave(a)gretlml.univpm.it
Website: https://gretlml.univpm.it/postorius/lists/gretl-users.gretlml.univpm.it/
3:44 p.m.
Am 16.05.2020 um 06:59 schrieb Burak Korkusuz:
Sorry, my fault, thanks for responding. I am adding my data to the
email. I have 5-minute realised variance data for FTSE.
Basically, I am trying to do rolling windows forecasting using the
HAR-RV model (in time series). I will have the initial sample;
observation 24-336 (I labelled observation date as numbers).
Observations between 337-437 will be the out-of-sample. First, I want
to estimate HAR model with 24-336 obs. and then forecast
one-step-ahead forecast [337]. Next, estimate HAR with 25-337 obs. and
then forecast [338]. Estimate with 26-338 and then forecast [339] ...
I am not sure I am doing right with my codes and in my case rolling
windows forecasting is; one-step-ahead forecasted values are generated
by each sliding different HAR model (i.e. 24-336, 25-337, 26-338 ...).
I am doing that with HAR-RV model and later on, I will add to the
model as an exogenous variable –VIX, EPU, Industrial production etc.–
and compare forecasting performance.
OK, for convenience let me paste your code from the attached mycode.inp
here again:
<hansl>
set verbose off
loop i=1..100 -q
smpl 23+i 335+i
ols RV5_FTSE const RV5_FTSE(-1) HARWEEK HARMONTH
fcast 336+i 336+i 1 forecasted --recursive
endloop
</hansl>
Basically I'd say this looks fine. However, two or three comments:
1) At first glance it appears that HARWEEK and HARMONTH are
contemporaneous variables. If that's true, it's not a real forecasting
equation. But perhaps in the background those variables actually refer
to the past.
2) The '--recursive' option is redundant here, since you're only
specifying a single obs to be forecast -- apart from the fact that you
don't want a recursively estimated model (where the starting obs of the
estimation window is fixed).
cheers
sven
7:40 p.m.
Yes, HARWEEK and HARMONTH variables refer to the past and they are already averaged
weekly/monthly values. The thing is as a series I cannot pick up one-step-ahead forecasted
values that is generated by re-estimated regressions. When I run, the codes generate a
series has only 436th forecasted observation.
The below codes give output as I would like to do (one-step-ahead rolling windows
forecasting) but I cannot save a series those one-step-ahead forecasted
values.<hansl>
set verbose off
loop i=1..100 -q
smpl 23+i 335+i
ols RV5_FTSE const RV5_FTSE(-1) HARWEEK HARMONTH
fcast 336+i 336+i
endloop
</hansl>
On Monday, May 18, 2020, 04:41:49 PM GMT+1, Sven Schreiber <svetosch(a)gmx.net>
wrote:
Am 16.05.2020 um 06:59 schrieb Burak Korkusuz:
Sorry, my fault, thanks for responding. I am adding my data to the email. I have 5-minute
realised variance data for FTSE.
Basically, I am trying to do rolling windows forecasting using the HAR-RV model (in time
series). I will have the initial sample; observation 24-336 (I labelled observation date
as numbers). Observations between 337-437 will be the out-of-sample. First, I want to
estimate HAR model with 24-336 obs. and then forecast one-step-ahead forecast [337]. Next,
estimate HAR with 25-337 obs. and then forecast [338]. Estimate with 26-338 and then
forecast [339] ...
I am not sure I am doing right with my codes and in my case rolling windows forecasting
is; one-step-ahead forecasted values are generated by each sliding different HAR model
(i.e. 24-336, 25-337, 26-338 ...). I am doing that with HAR-RV model and later on, I will
add to the model as an exogenous variable –VIX, EPU, Industrial production etc.– and
compare forecasting performance.
OK, for convenience let me paste your code from the attached mycode.inp here again:
<hansl>
set verbose off
loop i=1..100 -q
smpl 23+i 335+i
ols RV5_FTSE const RV5_FTSE(-1) HARWEEK HARMONTH
fcast 336+i 336+i 1 forecasted --recursive
endloop
</hansl>
Basically I'd say this looks fine. However, two or three comments:
1) At first glance it appears that HARWEEK and HARMONTH are contemporaneous variables. If
that's true, it's not a real forecasting equation. But perhaps in the background
those variables actually refer to the past.
2) The '--recursive' option is redundant here, since you're only specifying a
single obs to be forecast -- apart from the fact that you don't want a recursively
estimated model (where the starting obs of the estimation window is fixed).
cheers
sven
_______________________________________________
Gretl-users mailing list -- gretl-users(a)gretlml.univpm.it
To unsubscribe send an email to gretl-users-leave(a)gretlml.univpm.it
Website: https://gretlml.univpm.it/postorius/lists/gretl-users.gretlml.univpm.it/
9:33 p.m.
Am 18.05.20 um 21:40 schrieb Burak Korkusuz:
Yes, HARWEEK and HARMONTH variables refer to the past and they are
already averaged weekly/monthly values. The thing is as a series I
cannot pick up one-step-ahead forecasted values that is generated by
re-estimated regressions. When I run, the codes generate a series has
only 436th forecasted observation.
The below codes give output as I would like to do (one-step-ahead
rolling windows forecasting) but I cannot save a series those
one-step-ahead forecasted values.
<hansl>
set verbose off
loop i=1..100 -q
smpl 23+i 335+i
ols RV5_FTSE const RV5_FTSE(-1) HARWEEK HARMONTH
fcast 336+i 336+i
endloop
</hansl>
Well you can always use the $fcast accessor to get the result of the
last fcast command. So before the loop create a variable to be filled,
e.g. like this:
series pointforecast = NA
And then directly after the fcast line but still inside the loop you
insert something like:
pointforecast[336 + i] = $fcast
This should give you the wanted series of 1-step-ahead rolling forecasts.
cheers
sven
9:45 p.m.
ohh, this time ok, I got it, Thank you very much I really appreciate for your help.
On Monday, May 18, 2020, 10:33:33 PM GMT+1, Sven Schreiber <svetosch(a)gmx.net>
wrote:
Am 18.05.20 um 21:40 schrieb Burak Korkusuz:
Yes, HARWEEK and HARMONTH variables refer to the past and they are already averaged
weekly/monthly values. The thing is as a series I cannot pick up one-step-ahead forecasted
values that is generated by re-estimated regressions. When I run, the codes generate a
series has only 436th forecasted observation.
The below codes give output as I would like to do (one-step-ahead rolling windows
forecasting) but I cannot save a series those one-step-ahead forecasted values.
<hansl>
set verbose off
loop i=1..100 -q
smpl 23+i 335+i
ols RV5_FTSE const RV5_FTSE(-1) HARWEEK HARMONTH
fcast 336+i 336+i
endloop
</hansl>
Well you can always use the $fcast accessor to get the result of the last fcast command.
So before the loop create a variable to be filled, e.g. like this:
series pointforecast = NA
And then directly after the fcast line but still inside the loop you insert something
like:
pointforecast[336 + i] = $fcast
This should give you the wanted series of 1-step-ahead rolling forecasts.
cheers
sven
_______________________________________________
Gretl-users mailing list -- gretl-users(a)gretlml.univpm.it
To unsubscribe send an email to gretl-users-leave(a)gretlml.univpm.it
Website: https://gretlml.univpm.it/postorius/lists/gretl-users.gretlml.univpm.it/
1459
days inactive
1463
days old
6 comments
2 participants
participants (2)
-
Burak Korkusuz -
Sven Schreiber