Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: Rolling regression based on trading days


From   Nick Cox <[email protected]>
To   "[email protected]" <[email protected]>
Subject   Re: st: Rolling regression based on trading days
Date   Mon, 29 Apr 2013 10:08:04 +0100

It's me that is being slow. You did say that you have a business
calendar format in operation. So your last date is (presumably)
31dec2012 (not 31dec2002).

But I don't understand why the -xtset- is reporting ms, so I remain
unclear where the problem is.

Nick
[email protected]


On 29 April 2013 09:55, Bernice Mi <[email protected]> wrote:
> Hi Nick,
>
> Sorry for confusing you. My data look like this:
> company date price return
> 1      02jan1990
> 1      03jan1990
> 1      04jan1990
> .
> .
> .
> 1     31dec1991
> 2     02jan1980
> 2     03jan1980
> .
> .
> .
> 2     31dec2012
> .
> .
> .
> So, 8399 is the max period for one company say starting 02jan1980 - 31dec2002.
>
>
> Also, "describe date" gives us:
>
>               storage  display     value
> variable name   type   format      label      variable label
> ----------------------------------------------------------------------------------------------------------------------------------------------
> date            float  %tbbcal
>
> where "bcal" is the business calendar I have defined.
>
> Many thanks for your patience Nick!!
>
> On Mon, Apr 29, 2013 at 6:42 PM, Nick Cox <[email protected]> wrote:
>> Your -date- variable varies between 0 and 8399. That can't be a daily
>> date variable that includes dates like 31 Dec 2012.
>>
>> . di %d 8399
>> 30dec1982
>>
>> . di %d 0
>> 01jan1960
>>
>> So what you are showing us? You are -list-ing something; what is it?
>>
>> I also asked for -describe date-.
>> Nick
>> [email protected]
>>
>>
>> On 29 April 2013 09:25, Bernice Mi <[email protected]> wrote:
>>> Many thanks for your prompt reply Nick. Although delta is .001
>>> seconds, when I run rolling regression, it is still based on calendar
>>> days (not trading days as I want).
>>>
>>> The partial results are below. I really appreciate your help!
>>> 864761. | 26nov2012 |
>>> 864762. | 27nov2012 |
>>> 864763. | 28nov2012 |
>>> 864764. | 29nov2012 |
>>> 864765. | 30nov2012 |
>>>         |-----------|
>>> 864766. | 03dec2012 |
>>> 864767. | 04dec2012 |
>>> 864768. | 05dec2012 |
>>> 864769. | 06dec2012 |
>>> 864770. | 07dec2012 |
>>>         |-----------|
>>> 864771. | 10dec2012 |
>>> 864772. | 11dec2012 |
>>> 864773. | 12dec2012 |
>>> 864774. | 13dec2012 |
>>> 864775. | 14dec2012 |
>>>         |-----------|
>>> 864776. | 17dec2012 |
>>> 864777. | 18dec2012 |
>>> 864778. | 19dec2012 |
>>> 864779. | 20dec2012 |
>>> 864780. | 21dec2012 |
>>>         |-----------|
>>> 864781. | 24dec2012 |
>>> 864782. | 26dec2012 |
>>> 864783. | 27dec2012 |
>>> 864784. | 28dec2012 |
>>> 864785. | 31dec2012 |
>>>         +-----------+
>>>
>>> .
>>> end of do-file
>>>
>>> . do "C:\Users\lmi\AppData\Local\Temp\STD02000000.tmp"
>>>
>>> . su date
>>>
>>>     Variable |       Obs        Mean    Std. Dev.       Min        Max
>>> -------------+--------------------------------------------------------
>>>         date |    864785    4144.932    2229.524          0       8399
>>>
>>> .
>>>
>>> On Mon, Apr 29, 2013 at 6:16 PM, Nick Cox <[email protected]> wrote:
>>>> It seems that you have daily data. So why does -xtset- report milliseconds?
>>>>
>>>> Show us the results of say
>>>>
>>>> . describe date
>>>> . su date
>>>> . list date in 1/5
>>>>
>>>> I don't know offhand whether -rolling- is smart about business
>>>> calendars. Introducing business calendars is an on-going project for
>>>> StataCorp and I don't have access to Stata 12 right now.
>>>>
>>>> But if Stata thinks your time variable is clock time in ms, it won't
>>>> be paying much attention to business calendars. As far as it is
>>>> concerned you have in total about 1 second's worth of data.
>>>>
>>>> Nick
>>>> [email protected]
>>>>
>>>>
>>>> On 29 April 2013 08:53, Bernice Mi <[email protected]> wrote:
>>>>> Hi All,
>>>>>
>>>>> I have an unbalanced panel data with the dates being trading days. I
>>>>> have already defined and formatted (using %tbcalname) the dates as
>>>>> business calendar. I checked by testing the lead and lag days - it
>>>>> worked. However, when I xtset company date, it shows that delta is
>>>>> 0.001 seconds. Moreover, when I run rolling regression for 3 years, it
>>>>> turns up to be 1035 (roughly 360*3) replications not around 780 (i.e.
>>>>> 260*3) replications.
>>>>>
>>>>> I am new user of Stata and could not figure out where I went wrong.
>>>>> Please help me.
>>>>>
>>>>> Many thanks!!
>>>>>
>>>>> Bernice
>>>>> *
>>>>> *   For searches and help try:
>>>>> *   http://www.stata.com/help.cgi?search
>>>>> *   http://www.stata.com/support/faqs/resources/statalist-faq/
>>>>> *   http://www.ats.ucla.edu/stat/stata/
>>>> *
>>>> *   For searches and help try:
>>>> *   http://www.stata.com/help.cgi?search
>>>> *   http://www.stata.com/support/faqs/resources/statalist-faq/
>>>> *   http://www.ats.ucla.edu/stat/stata/
>>> *
>>> *   For searches and help try:
>>> *   http://www.stata.com/help.cgi?search
>>> *   http://www.stata.com/support/faqs/resources/statalist-faq/
>>> *   http://www.ats.ucla.edu/stat/stata/
>> *
>> *   For searches and help try:
>> *   http://www.stata.com/help.cgi?search
>> *   http://www.stata.com/support/faqs/resources/statalist-faq/
>> *   http://www.ats.ucla.edu/stat/stata/
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/faqs/resources/statalist-faq/
> *   http://www.ats.ucla.edu/stat/stata/
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/


© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index