GARCH

Research on Safe Withdrawal Rates

Moderator: hocus2004

Post Reply
User avatar
gummy
*** Veteran
Posts: 107
Joined: Wed May 14, 2003 4:00 am
Location: Hogsville, ON
Contact:

GARCH

Post by gummy »

The 2003 Nobel prize in economics was awarded for (among other things) GARCH, a mechanism for predicting future volatility, based upon recent historical data.

If somebuddy says GARCH, one is tempted to respond "Bless you" :D

Anyway, GARCH (which stands for Autoregressive Conditional Heteroskedasticity) obviously has applications to SWR (among other things).

Has anybuddy here looked at it?

I've been struggling with it for two weeks
JWR1945
***** Legend
Posts: 1697
Joined: Tue Nov 26, 2002 3:59 am
Location: Crestview, Florida

Post by JWR1945 »

These are some remarks by Benoit Mandelbrot. They are from pages 241-242 of The {Mis}behavior of Markets.
We mathematicians and physicists love what we call invariance...Fractal geometry is the mathematics of one such invariance in the physical world...Statisticians have a kindred concept, called stationarity: A stationary time series has the same basic statistical properties throughout. Economists argue their field may be different....Following that thinking, many recent models of price variation try to explain the obviously shifting pattern of volatility by inserting parameters that change by the day, hour, and second; such are in the GARCH family mentioned earlier....I would rather not dismiss the existence of invariances but continually look for them hiding in non-obvious places. Invariances make life easier....My multifractal model works with just such a set of consistent parameters.
Have fun.

John R.

BTW: The G in GARCH stands for Generalized.
JWR1945
***** Legend
Posts: 1697
Joined: Tue Nov 26, 2002 3:59 am
Location: Crestview, Florida

Post by JWR1945 »


User avatar
gummy
*** Veteran
Posts: 107
Joined: Wed May 14, 2003 4:00 am
Location: Hogsville, ON
Contact:

Post by gummy »

OOPS!
I should have said the 2003 Nobel prize was awarded for ARCH.
The G came years later :?

Try as I might, I can't find any useful application of ARCH (or GARCH).
Most of the literature seems to be replete with mathematical machinations.
(I have my own simple-minded ritual to incorporate volatility clustering.)

One-of-these-days I'll see how it affects Monte Carlo predictions for SWR
:roll:
Mike
*** Veteran
Posts: 278
Joined: Sun Jul 06, 2003 4:00 am

Post by Mike »

predicting future volatility, based upon recent historical data
This would suggest to me short term momentum, or at least non random volatility.
hocus2004
Moderator
Posts: 752
Joined: Thu Jun 10, 2004 7:33 am

Post by hocus2004 »

One-of-these-days I'll see how it affects Monte Carlo predictions for SWR

Gummy:

Have you given any thought to the idea of doing a tutorial at your web site explaining the findings that JWR1945 has come to with the research he has posted to this board? Such a tutorial would accomplish two important goals, in my view. One, it would put some people on notice of the true extent of the risk they are taking on in investing heavily in stocks at today's valuation levels and thereby might save a few retirements from going bust. Two, it might bring a few new faces to this board, making it a more powerful learning resource for all who make use of it.

I would be most grateful if you would give the ides some consideration.
User avatar
gummy
*** Veteran
Posts: 107
Joined: Wed May 14, 2003 4:00 am
Location: Hogsville, ON
Contact:

Post by gummy »

Mike:
The use of ARCH (or GARCH) is to modify (on a daily basis) the volatility of a stock in order to reflect current conditions.
That would include momentum and, certainly, dramatic changes in volatility.
Since the scheme involves random variables it wouldn't be non random.

hocus2004:
I'd be happy to stick in a link to some thread which discusses modifying SWR predictions based upon changing conditions.
Can you give me a URL?

Naturally, I like my sensible withdrawals strategy best :D

http://www.gummy-stuff.org/sensible_withdrawals.htm
hocus2004
Moderator
Posts: 752
Joined: Thu Jun 10, 2004 7:33 am

Post by hocus2004 »

I'd be happy to stick in a link to some thread which discusses modifying SWR predictions based upon changing conditions.

Thanks, Gummy. A link would be great.

I'd like to leave it for JWR1945 to pick the link that in his judgment best serves to advance the goals of this board community.
JWR1945
***** Legend
Posts: 1697
Joined: Tue Nov 26, 2002 3:59 am
Location: Crestview, Florida

Post by JWR1945 »

The best link might be to bpp's graphs at the bottom of the SWR Research Section. It is obvious that something is going on. Valuations matter. P/E10 tells us a lot about withdrawal rates and safety.
http://www.nofeeboards.com/jwr/jwr.html

Two sticky posts at the top of this page are probably the best additional references. They are designed for newcomers. They have links that fill in details.
May 2004 Overview dated Wed May 19, 2004.
http://nofeeboards.com/boards/viewtopic.php?t=2505
General Guidance dated Mon Apr 26, 2004.
http://nofeeboards.com/boards/viewtopic.php?t=2420

It is usually a good idea to consider entire threads, not simply the starting posts.

Finally, we need to mention that we have started using the phrase Historical Surviving Withdrawal Rates instead of Historical Database Rates. We need to point out that we define Safe Withdrawal Rates as being the lower confidence limit associated with our calculated rates. The upper confidence limit is defines a High Risk Withdrawal Rate.

Have fun.

John R.
User avatar
gummy
*** Veteran
Posts: 107
Joined: Wed May 14, 2003 4:00 am
Location: Hogsville, ON
Contact:

Post by gummy »

John:
For years (while writing hundreds of tutorials) I've often wished I could ask the authors (of the various articles) if they would write a tutorial that the simple-minded investor could understand (without the jargon).

However, lacking the guts to make such a request, I struggled with the dozens of webpages describing some financial idea associated with each tutorial.
(For example, I've downloaded and read over two dozen papers on ARCH/GARCH and struggled for over two weeks ... some fifty hours.)

Aah, but now I have the author .. and the guts! :)

Rather than struggling thru' your references...

what's HDBR80? The Historical Data Base Rate for an 80 year withdrawal?
what's are the "Two sticky posts at the top of this page"?
is it true that there are over two dozen threads involved in the various links to links to links that you mention?
...


... would you consider writing a tutorial on all that stuff?

http://www.gummy-stuff.org/JWR.htm

I'd be happy to put it on my website (along with any links you provide in the tutorial).
User avatar
gummy
*** Veteran
Posts: 107
Joined: Wed May 14, 2003 4:00 am
Location: Hogsville, ON
Contact:

Post by gummy »

P.S.
If you belive that the SWR analysis is important, then it may (who knows?) get a wider audience:

http://www.nedstatbasic.net/s?tab=1&link=3&id=3014409
JWR1945
***** Legend
Posts: 1697
Joined: Tue Nov 26, 2002 3:59 am
Location: Crestview, Florida

Post by JWR1945 »

gummy wrote:Rather than struggling thru' your references...

what's HDBR80? The Historical Data Base Rate for an 80 year withdrawal?
what's are the "Two sticky posts at the top of this page"?
is it true that there are over two dozen threads involved in the various links to links to links that you mention?
...

... would you consider writing a tutorial on all that stuff?

http://www.gummy-stuff.org/JWR.htm

I'd be happy to put it on my website (along with any links you provide in the tutorial).
HDBR50 and HDBR80 are standard portfolios that we use for a baseline.

HDBR50 consists of 50% stocks and 50% commercial paper. It is rebalanced annually. Expenses are set at 0.20% of the current balance of the portfolio.

HDBR80 consists of 80% stocks and 20% commercial paper. It is rebalanced annually. Expenses are set at 0.20% of the current balance of the portfolio.

When I use the notation HDBR50 and HDBR80, I am usually restricting myself to looking at survival rates over a 30-year period.

The two sticky posts are those that I provided links for:
May 2004 Overview dated Wed May 19, 2004.
General Guidance dated Mon Apr 26, 2004.
is it true that there are over two dozen threads involved in the various links to links to links that you mention?
Yes, it is true. Whether one reads them depends upon one's interest in them. They provide an efficient way to gain an understanding of much of the research at this board.

I am willing to put something together. Some guidance will be helpful. I am not going to come close to matching your presentation style and its entertainment value.

Have fun.

John R.
hocus2004
Moderator
Posts: 752
Joined: Thu Jun 10, 2004 7:33 am

Post by hocus2004 »

Would you consider writing a tutorial on all that stuff? ...I'd be happy to put it on my website (along with any links you provide in the tutorial).

I am very grateful for the wonderful opportunity that you are offering us, gummy.

My intent in suggeting that you pick the link, JWR1945, was not to pawn the work off on you. I didn't think that it should be me making the choice since it is your research that is the engine driving this machine. If I can offer any help in organizing material or in writing up background or whatever, I want to do that.

My sense is that gummy is now offering something more than just a link. He is using the word "tutorial." That indeed does mean (as he suggests in his "p.s." post) that we have an opportunity to reach some people with our message. I'll put forward here my thoughts as to how we should proceed. If anyone in the community (especially JWR1945 and gummy, but really anyone who has a suggestion to put forward that he or she thinks might be helpful), has thoughts contrary to mine, please just let me know and we will make whatever adjustments are necessary to attain a community consensus.

My thought is that I should take the lead on preparing a draft tutotial. I have never written in tutorial form before, but I will take a look at the other stuff at gummy's site and aim to follow that format at least to some extent. I will use the links offered above by JWR1945 as my starting point, but will go beyond them if I feel a need to do so to serve the purpose served by a tutorial. I will seek assistance from JWR1945 on the technical sorts of questions that I am not capable of handling on my own (just about anything which requires references to numbers or to difficult concepts like addition or subtraction or stuff of that nature).

It will take me some time to do this work as I have lots of other items on my plate. But I don't see that as a big problem; it is better to get this done right than it is to get it done fast. When I have something written, I will post it here and seek comment from the SWR Research Group board community. Only when we have a consensus here will I forward it to gummy. He will of course have complete discretion to make whatever changes he sees fit to make before posting the tutorial to his site.

Does that approach make sense?
User avatar
gummy
*** Veteran
Posts: 107
Joined: Wed May 14, 2003 4:00 am
Location: Hogsville, ON
Contact:

Post by gummy »

John, if you put together a post with the ideas and links involved (or e-mail me the text, if you'd like) , I'd be happy to put them into my "standard format" (with my %#$@!* sidekick who asks probing questions) with a note that they are based upon your NFB posts (or the text that you send me).

I'd send you drafts for your approval, of course.
We can work on it together.

e-mail: pjponzo@golden.net

Note:
According to e-mail that I receive, it's my sidekick who gets most of the credit for the entertainment value :?

I usually get: "Can't you dump the math and get to the point?"
hocus2004
Moderator
Posts: 752
Joined: Thu Jun 10, 2004 7:33 am

Post by hocus2004 »

I am willing to put something together.

Your response was not up when I began crafting my own, JWR1945. Either way of proceeding is OK with me. Please just let me know your preference.

My view is that I want our manner of proceeding to reflect the reality that it is your research that we are trying to get the word out on. You obviously possess a far better understanding of the technical side of things than I do. But in the event that there is anything that I can do on the writing side to help out a bit, I want to make it clear that I am happy to do so.
User avatar
gummy
*** Veteran
Posts: 107
Joined: Wed May 14, 2003 4:00 am
Location: Hogsville, ON
Contact:

Post by gummy »

Hey hocus!
I didn't mean to leave y'all out of this :o

I just have so little knowlege of what the analysis has been (over several years, I understand) and I'm so old and decrepit that reading all the stuff would shorten my life by x years ... or is it N years?

Anyway, I'd be happy to wait for whatever info I get ... from whomever :lol:

Currently, I'm now interested in (gasp!) S-video and HDTV and all that stuff (about which I knew nothing):
http://www.gummy-stuff.org/TV-stuff.htm

P.S.
You may ask "If you spent fifty hours on GARCH why not fifty on this SWR stuff?"

However, last time I looked, I have a couple of dozen tutorials on SWR stuff and HDTV is more interesting ...at this time ;)
hocus2004
Moderator
Posts: 752
Joined: Thu Jun 10, 2004 7:33 am

Post by hocus2004 »

The approach suggested in the most recent gummy post also makes a lot of sense. The two of you are both Numbers Guys and I am not, so the two of you understand what the other is saying in a way that I do not. Please don't feel a need to include me in any way, shape, or form. I'm perfectly OK with being left out of the loop on this.

All that I am trying to say is that it is not right for me to put an idea on the table and then run into the other room when it is time to divide up the work tasks. If it is possible for me to avoid all the work in good conscience, I am totally cool with that.
hocus2004
Moderator
Posts: 752
Joined: Thu Jun 10, 2004 7:33 am

Post by hocus2004 »

Hey hocus!
I didn't mean to leave y'all out of this


Uh-oh.

I thought I was on a roll there for a little bit.
User avatar
gummy
*** Veteran
Posts: 107
Joined: Wed May 14, 2003 4:00 am
Location: Hogsville, ON
Contact:

Post by gummy »

Mamma mia!
Posts at 11:25 then 11:31 then 11:33 then 11:37 then 11:39 then 11:43 then ...

One refers to "the most recent post"
then there are a couple more in between
Mike
*** Veteran
Posts: 278
Joined: Sun Jul 06, 2003 4:00 am

Post by Mike »

Thanks Gummy.
Post Reply