COMPGROUPS.NET | Search | Post Question | Groups | Stream | About | Register

### DCC GARCH - UCSD Garch tool box

• Email
• Follow

I tried to obtain the results of dcc garch by running the file dcc_mvgarch.m with the stock market index data. but of no use!! After reading the code I found out that I need to do process the stock data. This is what I found about processing the data.

data          = A zero mean t by k vector of residuals from some filtration

I just do'nt understand how to can I convert my index data into zero mean vector of residuals and what sort of filtration am i supposed to use??

 0

See related articles to this posting

"frm cfa" wrote in message <igdls9\$31g\$1@fred.mathworks.com>...
> I tried to obtain the results of dcc garch by running the file dcc_mvgarch.m with the stock market index data. but of no use!! After reading the code I found out that I need to do process the stock data. This is what I found about processing the data.
>
> data          = A zero mean t by k vector of residuals from some filtration
>
> I just do'nt understand how to can I convert my index data into zero mean vector of residuals and what sort of filtration am i supposed to use??
>

I'm not an expert on the toolbox (or the Matlab behind it), but in finance terms, you probably want to do something like this:

1. Derive a return series from the index series. The index is nonstationary, returns generally are.

2. Estimate a return model on the returns. (That's what I think is meant by filtration in this context.) The goal is to derive residuals 'et' for each return, being the difference between the observed return and the estimated return, such that 'et' has a zero mean.

At the very simplest level, that could simply be calculating the mean and deducting it. (Your model for each asset is then rt = m + et, where rt are your returns, m is their mean and et is the data series.) If you want something more complex, you might fit an ARMA model or such like.

The Ucsd Garch tool box was developed by Kevin Sheppard (contact details within the .m file). If you have any further problems, he may be able to help you get it working if you approach him directly.

I hope that helps.
 0

Million thanks William, this is really very useful and helpful!!
 0

Did you ever get it to work? I've been having the same problem... thanks
 0

"Ross " <inthepinktrading@gmail.com> wrote in message <ijaj57\$qs4\$1@fred.mathworks.com>...
> Did you ever get it to work? I've been having the same problem... thanks

try the price2ret() function. The function converts price time series to a return series via ret=ln(price_t/price_t-1). This should only be used for high-frequency data (daily)
 0
Reply gmr2 (3) 6/22/2011 10:59:04 PM

"frm cfa" wrote in message <igdls9\$31g\$1@fred.mathworks.com>...
> I tried to obtain the results of dcc garch by running the file dcc_mvgarch.m with the stock market index data. but of no use!! After reading the code I found out that I need to do process the stock data. This is what I found about processing the data.
>
> data          = A zero mean t by k vector of residuals from some filtration
>
> I just do'nt understand how to can I convert my index data into zero mean vector of residuals and what sort of filtration am i supposed to use??
>

Thanks
 0

"frm cfa" wrote in message <igdls9\$31g\$1@fred.mathworks.com>...
> I tried to obtain the results of dcc garch by running the file dcc_mvgarch.m with the stock market index data. but of no use!! After reading the code I found out that I need to do process the stock data. This is what I found about processing the data.
>
> data          = A zero mean t by k vector of residuals from some filtration
>
> I just do'nt understand how to can I convert my index data into zero mean vector of residuals and what sort of filtration am i supposed to use??
>

Hi, Where did you find this file?

Thanks
Jen
 0
Reply aj_cutie99 (4) 9/14/2011 12:36:10 PM

"Jen" wrote in message <j4q73q\$9st\$1@newscl01ah.mathworks.com>...
> "frm cfa" wrote in message <igdls9\$31g\$1@fred.mathworks.com>...
> > I tried to obtain the results of dcc garch by running the file dcc_mvgarch.m with the stock market index data. but of no use!! After reading the code I found out that I need to do process the stock data. This is what I found about processing the data.
> >
> > data          = A zero mean t by k vector of residuals from some filtration
> >
> > I just do'nt understand how to can I convert my index data into zero mean vector of residuals and what sort of filtration am i supposed to use??
> >
>
> Hi, Where did you find this file?
>
> Thanks
> Jen

after installing the Ucsd_garch toolbox correctly, the dcc_mvgarch.m file can be found under (default) installation folder
C:\Program Files\MATLAB\R2011b\toolbox\Ucsd_garch\MV Garch\Correlation MVGarch
 0
Reply thisisafake (7) 2/12/2012 7:29:10 PM

"frm cfa" wrote in message <igdls9\$31g\$1@fred.mathworks.com>...
> I tried to obtain the results of dcc garch by running the file dcc_mvgarch.m with the stock market index data. but of no use!! After reading the code I found out that I need to do process the stock data. This is what I found about processing the data.
>
> data          = A zero mean t by k vector of residuals from some filtration
>
> I just do'nt understand how to can I convert my index data into zero mean vector of residuals and what sort of filtration am i supposed to use??
>

I have completed my thesis using the UCSD Toolbox. All the Matlab code works. My work is available for others to advance with. I hope others find it useful.
 0
Reply thisisafake (7) 5/22/2012 2:34:07 PM

"Garth Mortensen" <thisisafake@email.com> wrote in message
news:jpg84v\$gtu\$1@newscl01ah.mathworks.com...
> "frm cfa" wrote in message <igdls9\$31g\$1@fred.mathworks.com>...
>> I tried to obtain the results of dcc garch by running the file
>> dcc_mvgarch.m with the stock market index data. but of no use!! After
>> reading the code I found out that I need to do process the stock data.
>> This is what I found about processing the data.
>>
>> data          = A zero mean t by k vector of residuals from some
>> filtration
>>
>> I just do'nt understand how to can I convert my index data into zero mean
>> vector of residuals and what sort of filtration am i supposed to use??
>>
>
> I have completed my thesis using the UCSD Toolbox. All the Matlab code
> works. My work is available for others to advance with. I hope others find
> it useful.

Congratulations!

--
Steve Lord
slord@mathworks.com
http://www.mathworks.com

 0
Reply slord (13686) 5/22/2012 2:54:21 PM

My thesis applied MVGARCH to two return series. I put all my work and code online so others can benefit.

I hope it helps.
 0
Reply thisisafake (7) 9/28/2012 6:06:09 PM

10 Replies
1635 Views

Similar Articles

12/8/2013 3:10:39 AM
page loaded in 20693 ms. (0)