Good work by jwr1945

Financial Independence/Retire Early -- Learn How!
Post Reply
User avatar
ataloss
**** Heavy Hitter
Posts: 559
Joined: Mon Nov 25, 2002 3:00 am

Good work by jwr1945

Post by ataloss » Fri Oct 31, 2003 9:36 am

I was a little concerned by this statement:
I strongly recommend using the years after 1920 for projecting safe withdrawal rates. We live in an era with the Federal Reserve and without a gold standard. That is sufficient to favor the later time period.


http://nofeeboards.com/boards/viewtopic ... 8929#p8929

but I see that in more recent posts jwr1945 has been using the full data set from 1871. Excluding data that isn't "well behaved" seems undesirable to me so I am glad to see this development.
Have fun.

Ataloss

wanderer
*** Veteran
Posts: 363
Joined: Tue Nov 26, 2002 9:33 am
Location: anytown, usa

Post by wanderer » Fri Oct 31, 2003 5:17 pm

I'm still trobled by his use of phrases like "we know" "we've proven" wrt safe withdrawal rates and the mean, standard deviation, etc. of the 'true' SWR at any given point in time. Is he still using those phrases? 3 or 4 non-overlapping data points don't constitute much 'proof' in a data-mining exercise like this, IMO.
regards,

wanderer

The field has eyes / the wood has ears / I will see / be silent and hear

peteyperson
**** Heavy Hitter
Posts: 525
Joined: Tue Nov 26, 2002 6:46 am

Post by peteyperson » Fri Oct 31, 2003 5:32 pm

I have to agree with you, wanderer. I too have been troubled with this. Financial writers tend to put forward their own ideas and conclusions based on their research but rarely do they state categorically. It is usually up to the reader to determine their beliefs and investment approach based on what they have read and accepted as fact, rather than what has been proclaimed by the writer as such. It's not a science with many right or wrong answers, once you're in the right ballpark opinions vary dramatically. (I think JWR gets egged on by hocus and gets over excited as they run energetically down a blind alley leading nowhere).

As a fixed swr analysis seems fundamentally flawed. Everyone has a different asset mix and behaviour during downturns that a standard consistent approach. Psychology, fear of further losses, acting in favor of short-term losses vs long-term greater returns by holding on etc affect portfolio management during FIRE. A fixed approach & w.d. rate for all seems to be trying to bang a nail into the wrong hole. To stretch the analogy out further, I'm not even sure there is a correct hole.

I recently posted a thread discussing portfolios structured from the position of protection and funding living standards during a downmarket vs. aggressively into equities & growth ignoring the risks of a long drawn out downmarket, but sadly no one joined in.

http://nofeeboards.com/boards/viewtopic.php?t=1588

Petey
wanderer wrote: I'm still troubled by his use of phrases like "we know" "we've proven" wrt safe withdrawal rates and the mean, standard deviation, etc. of the 'true' SWR at any given point in time. Is he still using those phrases? 3 or 4 non-overlapping data points don't constitute much 'proof' in a data-mining exercise like this, IMO.

wanderer
*** Veteran
Posts: 363
Joined: Tue Nov 26, 2002 9:33 am
Location: anytown, usa

Post by wanderer » Fri Oct 31, 2003 5:41 pm

I think JWR gets egged on by hocus

I agree. I'm sticking with the retired math prof.
regards,

wanderer

The field has eyes / the wood has ears / I will see / be silent and hear

User avatar
ataloss
**** Heavy Hitter
Posts: 559
Joined: Mon Nov 25, 2002 3:00 am

Post by ataloss » Fri Oct 31, 2003 5:41 pm

So you are saying that jwr has developed an approach based on past data that would have worked had we known about it?

I notice that jwr is careful not to call this a swr (since it is based on past data and might not be a "swr as swr is defined for the purposes of swr analysis")

still I am happy that he isn't excluding the data that didn't fit so well
Have fun.

Ataloss

wanderer
*** Veteran
Posts: 363
Joined: Tue Nov 26, 2002 9:33 am
Location: anytown, usa

Post by wanderer » Fri Oct 31, 2003 5:44 pm

It's real hard to conduct an out of sample test when each data point requires 30 years of data. It may not be as difficult after the year 3200 AD/CE.
regards,

wanderer

The field has eyes / the wood has ears / I will see / be silent and hear

raddr
*** Veteran
Posts: 265
Joined: Mon Nov 25, 2002 3:25 am
Contact:

Post by raddr » Sat Nov 01, 2003 4:05 am

wanderer wrote: It's real hard to conduct an out of sample test when each data point requires 30 years of data. It may not be as difficult after the year 3200 AD/CE.


Exactly. That's why some sort of simulation technique seems almost mandatory to fine tune the process and serve as a reality check. 130 years of data (the first 50-60 of which are of questionable quality) from an "index" which was not even in existence till the late 50's is pitifully little to work with. Very few people seem to grasp this concept, at least at the REHP. We're a bit more enlightened over here. :wink:

Post Reply