negative coefficients of ARCH terms in MGARCH-BEKK

Discussions of ARCH, GARCH, and related models
TomDoan
Posts: 7814
Joined: Wed Nov 01, 2006 4:36 pm

Re: negative coefficients of ARCH terms in MGARCH-BEKK

Unread post by TomDoan »

There's no way I could tell from that. Have you even graphed your data? You should know whether there are issues even before you fit a model.
prashantj
Posts: 88
Joined: Sun Apr 11, 2010 2:56 am

Re: negative coefficients of ARCH terms in MGARCH-BEKK

Unread post by prashantj »

This is what I got there are some outliers, but these are what I think we find with GARCH series?
Graph.RGF
(17.54 KiB) Downloaded 602 times
prashantj
Posts: 88
Joined: Sun Apr 11, 2010 2:56 am

Re: negative coefficients of ARCH terms in MGARCH-BEKK

Unread post by prashantj »

Finally, I think I have got the following model with my intraday data after adjusting for outliers, took a long! . I hope that is a good model, Tom?
SPill.jpg
SPill.jpg (235.52 KiB) Viewed 11022 times
Last edited by prashantj on Sun Nov 29, 2020 11:50 am, edited 1 time in total.
TomDoan
Posts: 7814
Joined: Wed Nov 01, 2006 4:36 pm

Re: negative coefficients of ARCH terms in MGARCH-BEKK

Unread post by TomDoan »

That looks more reasonable.
TomDoan
Posts: 7814
Joined: Wed Nov 01, 2006 4:36 pm

Re: negative coefficients of ARCH terms in MGARCH-BEKK

Unread post by TomDoan »

prashantj wrote:This is what I got there are some outliers, but these are what I think we find with GARCH series?
Graph.RGF
The series is black seems to have a quite a few big outliers of opposite signs in close proximity (possibly adjacent entries) with no apparent effect on volatility subsequent to that. That tends to be a sign of either a one data point typo in the price series or some other move and immediate correction that is seen in the market as an obvious fix.
prashantj
Posts: 88
Joined: Sun Apr 11, 2010 2:56 am

Re: negative coefficients of ARCH terms in MGARCH-BEKK

Unread post by prashantj »

Thanks , Tom. I appreciate it. I seen quite great features in RATS10. The computational power has improved significantly much faster than older version 8.2 I had. RATS saves a significance time in learning and looking for program like using open source softwares. And your help makes RATS unique. When the RATS11 you plan to release? I will be the first to get it.
Thanks!
Prashant
prashantj
Posts: 88
Joined: Sun Apr 11, 2010 2:56 am

Re: negative coefficients of ARCH terms in MGARCH-BEKK

Unread post by prashantj »

I was estimating BEKK MGARCH model by creating variance dummy using the procedure discussed in GARCH course workbook in Section 6.3 and I got the following output (am attaching truncated). I would like to get value for march(1,2). Is it possible to generate?

garch(mv=BEKK,rvectors=rd,hmatrices=hh,pmethod=simplex,asymmetric,xreg,stdresids=rstd,piters=20,iters=500,dist=T) / rsp rbg
# march19
dummy.jpg
dummy.jpg (41.9 KiB) Viewed 11008 times
TomDoan
Posts: 7814
Joined: Wed Nov 01, 2006 4:36 pm

Re: negative coefficients of ARCH terms in MGARCH-BEKK

Unread post by TomDoan »

It's a lower triangular matrix (has to be---a full 2x2 matrix wouldn't be identified) so the 1,2 element is zero.
prashantj
Posts: 88
Joined: Sun Apr 11, 2010 2:56 am

Re: negative coefficients of ARCH terms in MGARCH-BEKK

Unread post by prashantj »

got it. Thanks.
Prashant
prashantj
Posts: 88
Joined: Sun Apr 11, 2010 2:56 am

Re: negative coefficients of ARCH terms in MGARCH-BEKK

Unread post by prashantj »

Is till could not import intraday data with the time stamps. I tried all possibilies I can but of no avail. can you please help me with it to read the data in RATS10? I will send you the file.
Looking forward for your reply,
Prashant
Post Reply