Does VII work?

Rob Bennett moderates this Early Retirement Board
Abusive posting will not be tolerated.

Moderator: Founder of Hocomania

Does VII work?

Postby nfs » 02/12/08 at 17:48:13

I'm putting this here because all the other forums on this board are just for poking Rob and I'm aiming to do some darned serious work for a few minutes.

My object was to examine Rob's compendium of claims that VII is a superior investment strategy.  Since I needed a set of actionable guidelines rather more precise than the usual cloudy handwaving, I turned to his site and in particular this how-to article on VII.  I don't know how many times that's been changed since first written, and it's got its own internal contradictions, but I did the best I could to interpret Rob's instructions.

I distilled it down to the following rules:

1. Use Shiller's PE10 as a valuation guide.
2. You should own more stocks when PE10 is low and fewer when it's high.  Rob provides a few pegs to anchor the percentage of stocks, so I ended up with a scale that ran from 100% stocks at PE10 = 5 down to 0% stocks at PE10=35.
3. Minor changes don't matter, so the stock allocation is stepped by 10% at PE10 intervals of 3.  As an example, Rob fixes 30% stocks as reasonable at PE10=26, so we stick with that until PE10=29 when stocks would be reduced to 20% or until PE10=23 when they would be increased to 40%.
4. Rebalance if you get too far out of line, even if PE10 hasn't moved enough.  I used a 5% variation as the limit, so if the target is 30% stocks, you do nothing unless you're under 25% or over 35%.

I then ran a simulation on monthly data, using Shiller's own PE10 records with Ken French's monthly returns for both the risk-free rate and the market return.  (French's "market" is wider than the S&P 500 because his data is used to test the 3 factor model but it should be fine for this purpose.)  French's data goes back to mid 1926, so there's about 80 years of monthly data available.

The VII algorithm ends up with 150 rebalancing transactions over 80 years, so it's moving about twice a year.  While this could not have been done in early days because of transaction costs, it's no problem nowadays with the availability of inexpensive index funds that would cover the asset classes.

Against VII, I set the bogey of B&H.  The average PE10 since 1926 has been very close to 17 and, at PE10=17, VII would be 60% stocks, so the classic portfolio with 60% in stocks matches dead on.  The other 40% ends up in the risk free asset (in French's dataset, T-bills) so it's not a normal B&H portfolio (which would have bonds, not bills).  However, if VII is going to get stuck with bill rates when it's out of the market, then it seems fair that B&H does too.  B&H gets the same rebalancing rule as VII, adding to stocks if they get to 55% or below and reducing if they get to 65% or above.  B&H rebalances 58 times in 80 years.

I compared 10, 20, and 30 year annualized returns using these two methods and I'll post the graphs of one against the other in subsequent messages.  B&H is the black line and VII gets the lovely magenta.  The results are interesting.
All in all you're just another brick in the wall. - Roger Waters
nfs
***** God Member
 
Posts: 537
Joined: 03/14/05 at 13:21:11
Location: The Dry Side of the Wet Coast

Re: Does VII work?

Postby nfs » 02/12/08 at 17:53:35

Image
These are the (annualized) returns for the 30 year period ending on the month in question.  VII is pretty clearly superior, although the gap for the last 30 years (which means starting as long as 60 years ago) has been fairly small, sometimes vanishingly so, and appears to have narrowed more recently.
All in all you're just another brick in the wall. - Roger Waters
nfs
***** God Member
 
Posts: 537
Joined: 03/14/05 at 13:21:11
Location: The Dry Side of the Wet Coast

Re: Does VII work?

Postby nfs » 02/12/08 at 17:57:16

Image
This is annualized returns over 20 years.  Once again, VII looks generally better, although not very much so since the 1970s.
All in all you're just another brick in the wall. - Roger Waters
nfs
***** God Member
 
Posts: 537
Joined: 03/14/05 at 13:21:11
Location: The Dry Side of the Wet Coast

Re: Does VII work?

Postby nfs » 02/12/08 at 18:05:01

Image
These are 10 year annualized returns and now it's very nuanced.  If we had asked this sort of question back in 1996 with the then available data, it would look like a very noisy series, advantage generally to VII but not very much in favor for almost 40 years previous.

But then came the late 1990s and absolutely everything is turned upside down.  B&H wins by a mile, at least until 2002.

So, is VII really nonsense, or is devotion to B&H a consequence of recency bias?

(thus putting the cat squarely among the pigeons  ;D)
All in all you're just another brick in the wall. - Roger Waters
nfs
***** God Member
 
Posts: 537
Joined: 03/14/05 at 13:21:11
Location: The Dry Side of the Wet Coast

Re: Does VII work?

Postby Yipee-Ki-O » 02/12/08 at 18:21:03

Interesting. I'd be interested to see how VII fared against Coffeehouse over past 17 year period or vs. Vanguard Wellington for longer period.
Work is the refuge of those who have nothing better to do, Oscar Wilde
Yipee-Ki-O
***** God Member
 
Posts: 10553
Joined: 03/19/05 at 18:40:02

Re: Does VII work?

Postby DRiP_Guy » 02/12/08 at 19:14:17

Move over Wall Street Week Pundits, DripGuy has arrived!

I called it thusly:

If you were solving for the highest possible (albeit unlikely) theoretical return sequence in history (once it occurred), you would be hamstrung to the truth (which no one ever denied) that 'valuations matter.'

But, if you are solving for trying to obtain very near market returns in a REAL portfolio, over a long period of time, you simply accept that although valuations may well matter, the alternatives to handling that fact (lumping in, or DCA, or sitting on sidelines for a couple of decades) don't somehow favor wishing that PE10 was 7, and therefore pinning the success of your plan on waiting for a huge crash the likes of which has not been since since 1929. No one is saying you can't do that Rob, but don't be shocked when few follow your ideas, since there is not a scenario you can give where "Dodge, whinge, wince, hand wring, and wait?"  somehow manages to beat good old "buy-n-hold".
Robert Michael "Hocus" Bennett, October 22, 2010: "I have been living off my savings for 10 years"
User avatar
DRiP_Guy
***** God Member
 
Posts: 9148
Joined: 10/21/06 at 14:42:16

Re: Does VII work?

Postby Schroeder » 02/13/08 at 10:16:51

Excellent analysis, nfs. I can appreciate all the hard work that you put into this.

nfs wrote:The VII algorithm ends up with 150 rebalancing transactions over 80 years, so it's moving about twice a year. ?While this could not have been done in early days because of transaction costs, it's no problem nowadays with the availability of inexpensive index funds that would cover the asset classes.

This is a lot of transactions. It should give one pause to realize what VII entails. It's certainly not a tax-efficient strategy. After taxes, VII would likely trail buy-and-hold.

Schroeder
Proud Member of the Financial Freedom Community!
User avatar
Schroeder
***** God Member
 
Posts: 3892
Joined: 10/11/05 at 19:53:13

Re: Does VII work?

Postby sgeeeee » 02/13/08 at 21:19:34

Interesting stuff, but it doesn't really address SWR directly.  Remember that if you find an investment technique that provided higher returns 99 time out of 100, but that 100th time is the worst case scenario, then the SWR is reduced rather than improved.  

It is actually pretty easy to develop more complex investment algorithms designed to improve backtesting average returns or SWRs.  SWRs are determined by only a couple of periods in history (the Great Depression and the years of hyperinflation).  If you focus on these two periods and figure out what investment style worked best, then define a set of tests that uniquely identify both periods, you can easily come up with investment rules that backtest better than a simple fixed equity/bond split.  This kind of exercise is not likely to be very useful, however.  The rules you come up with are not likely to actually predict optimum performance in the futrue.  You can imagine rules like, "If the year ends with '29' take all your money out in cash at the beginning of the year, then invest heavily in the stock market two years later."  Such a rule would backtest well.  While most people would never consider a rule this blatantly targeting improved backtesting, rules more complex than simple allocation procedures are all likely to be misleading.  

I can't tell from the data what happens to SWR for this case.  But even if it improves, when I look at all of the data I am forced to conclude that VII is as likely to be worse than straight allocation in the future as it is to be better.  When you look at the historical record the appropriate conclusion is, "Valuation (as defined by PE10) matters . . . sometimes, but sometimes it doesn't."   :)
-sgeeeee
User avatar
sgeeeee
***** God Member
 
Posts: 719
Joined: 02/10/04 at 22:48:31
Location: Mesa, AZ

Re: Does VII work?

Postby hocus » 02/14/08 at 16:09:50

This is a lot of transactions. It should give one pause to realize what VII entails.

The easiest way to test VII is by using the Scenario Surfer. I've done hundreds of runs. It is rare for VII not to beat all three rebalancing strategies that the calculator tests for comparison purposes. VII is the winner about 90 percent of the time, 80 percent rebalancing is the winner on the remaining tests.

I usually make it a practice to engage in few transactions. I would guess that my average is about five over the 30-year time-period tested. There's no need to engage in lots of transactiions to make this work.

A very simple way to test it would be to go with 80 percent stocks for years with a starting P/E10 of up to 19 and with 20 percent stocks for years with a starting P/E10 of 19 and above. I am confident that VII would fare better in the majority of cases even though the approach is obviously ham-handed.

It of course makes sense that the numbers would not look nearly so good for tests covering only the years of the longest and strongest bull market in history. We will only know how those who invested heavily in stocks after prices got to la-la lands levels have done when there has been sufficient time for prices to return to normal levels again. Historically, investing heavily in stocks at the sorts of prices that apply today has never produced good long-term results. That's all we can say as of today.

Rob
Lying and sliming since May 12, 2002 --  Don't understand why intercst allows me to post here. Learn more about Rob Bennett, see link: http://retireearlyhomepage.com/bennett.html
User avatar
hocus
***** God Member
 
Posts: 5847
Joined: 02/11/04 at 12:31:16

Re: Does VII work?

Postby nfs » 02/14/08 at 16:20:06

hocus wrote:A very simple way to test it would be to go with 80 percent stocks for years with a starting P/E10 of up to 19 and with 20 percent stocks for years with a starting P/E10 of 19 and above. I am confident that VII would fare better in the majority of cases even though the approach is obviously ham-handed.

You might want to rethink the ham-handed approach.  The resulting graphs do not support your thesis very well.
All in all you're just another brick in the wall. - Roger Waters
nfs
***** God Member
 
Posts: 537
Joined: 03/14/05 at 13:21:11
Location: The Dry Side of the Wet Coast

Re: Does VII work?

Postby hocus » 02/14/08 at 17:42:40

You might want to rethink the ham-handed approach. ?The resulting graphs do not support your thesis very well.

Have you tried testing it on the Scenario Surfer, NFS? I haven't tested this, so I don't know for sure. My sense from the runs I have done, though, is that this would usually work.

An 80 percent allocation is usually a good allocation by the numbers. The approach I outlined would get you out of an 80 percent allocation only when stocks got to very high price levels. I don't say this is the best possible way to go; I doubt very much that it is. But it should produce good results.

Are your graphs focusing too much on what happened during the wild bull? Remember, we don't know what is going to happen to those who invested heavily in stocks at the super-high prices that have applied in recent years until prices return to more reaosnable levels. You need to look at the entire history of stocks, not just the history of a huge bull market (and not at a history that focuses excessively on that one time-period).

John and I are in the early stages of developing a calculator that would allow quick testing of these sorts of things (puttting the Surfer through 1,000 runs of a strategy rather than having to do one test at a time). When that calculator is ready, we will be able to test something like this quickly. For now, we need to test run by run, and that obviously takes more time.

I'd like to see lots of people get involved in testing different strategies. That sort of feedback would be helpful to lots of people in lots of ways.

Rob
Lying and sliming since May 12, 2002 --  Don't understand why intercst allows me to post here. Learn more about Rob Bennett, see link: http://retireearlyhomepage.com/bennett.html
User avatar
hocus
***** God Member
 
Posts: 5847
Joined: 02/11/04 at 12:31:16

Re: Does VII work?

Postby Yipee-Ki-O » 02/14/08 at 18:54:00

Sounds like a reasonable offer devoid of all that cotton candy appreciation that Rob so deplores. ;)
Work is the refuge of those who have nothing better to do, Oscar Wilde
Yipee-Ki-O
***** God Member
 
Posts: 10553
Joined: 03/19/05 at 18:40:02

Re: Does VII work?

Postby hocus » 02/14/08 at 19:14:58

Yeah, I'm hostile. ?You invited me to use a deck of marked cards.

The marked deck of cards is yours, NFS.

There have been hundreds of community members who have expressed a desire to explore the ideas behind the Valuation-Informed Indexing approach. There are lots and lots of threads with good information in them and there are a good number of articles at my site setting forth the insights that community members developed together. Valuation-Informed Indexing works.

Is it better suited for some circumstances than for others? Sure. Is it better suited for some types of investors than for others? Sure. Are there caveats that need to be stated? Sure. Are there questions we still need to explore? Sure.

All that is so. But the idea that there is some question at this point as to whether it works is an absurdity. If we hadn't proved that it worked beyond any reasonable doubt, why the heck would Greaney have spent six years of his life leading a Campaign of Terror to block interested parties from hearing about it?

It works for all who learn about it. If you want to benefit from the ideas developed by people you have smeared, you are free to do so. You need to get past the hostility to learn. That's the block.

The fact that you don't want it to work is your problem, it's not the problem of the approach or of?the ideas behind the approach or of the people who have helped develop those ideas and who have expressed a desire to learn more. You work through your issues, and all the good stuff is there for you and for all the others. I can answer questions about the investing ideas. I cannot help you with your personal issues of not being able to say the words "I" and "Was" and "Wrong."

I can say that it's a job that doesn't get easier by putting it off. Had Greaney said the magic words six years ago, every single community member (including him) would be in a better place today. When you get a number wrong, you need to fix it. That's the way it goes. I didn't make that rule. I just happen to live in a world in which that is the rule that applies.

Rob
Lying and sliming since May 12, 2002 --  Don't understand why intercst allows me to post here. Learn more about Rob Bennett, see link: http://retireearlyhomepage.com/bennett.html
User avatar
hocus
***** God Member
 
Posts: 5847
Joined: 02/11/04 at 12:31:16

Re: Does VII work?

Postby hocus » 02/14/08 at 19:18:08

You've got that passive-aggressive routine down pat.

It's called "civility."

All learning follows from tapping into its power.

Going back to the Grunt stage is not the key to successful long-term investing. I'm sure of it!

Rob
Lying and sliming since May 12, 2002 --  Don't understand why intercst allows me to post here. Learn more about Rob Bennett, see link: http://retireearlyhomepage.com/bennett.html
User avatar
hocus
***** God Member
 
Posts: 5847
Joined: 02/11/04 at 12:31:16

Re: Does VII work?

Postby nfs » 02/14/08 at 19:45:17

hocus wrote:Yeah, I'm hostile.  You invited me to use a deck of marked cards.

The marked deck of cards is yours, NFS.

No, sir, you are lying.  The graphs above in this thread were built from purely historical data, i.e. facts, and as good an implementation of your VII scheme as I could distill from your verbiage.  If I have misinterpreted something you wrote, you are cordially invited to correct it and I will rerun the simulations on the historical facts.  Your attempt above to replace the sliding scale implicit in your own blog entry with a ham-handed 80/20 was met with another graph, which any observer can quite clearly see works less well versus B&H than my original implementation.

The "marked deck" is the absolutely imaginary set of "returns" that is imbedded in your Scenario Surfer.  There is no historical basis for that set of "returns".  You made them up out of thin air.  The comment inadvertently left in JWR's spreadsheet documents it.  They do not come from any historical record.  Claim that they come from the historical record as you wished it would be if you like, but that's a different kettle of fish and all would agree when I say the fish is well past its sell-by date and rather malodorous.

If you'd like me to mark your deck, I'll be happy to do so.  I'll create an imaginary set of returns for you to imbed in the Scenario Surfer and then you can run your examples.  You will no doubt be surprised to discover that VII would never work in that instance.

The fact that you don't want it to work is your problem, it's not the problem of the approach or of the ideas behind the approach or of the people who have helped develop those ideas and who have expressed a desire to learn more. You work through your issues, and all the good stuff is there for you and for all the others.

You really are a dope.  Have you looked at the first few posts in this thread?  Have you looked at those graphs?  VII looks better than B&H.  Can you not just accept the gift with a little grace?

P.S.  The offer on the house will expire at 5 pm EST tomorrow.  I've got business out of town next week and wouldn't want to leave it hanging.  $230k.  If it's still for sale when I get back, I'll look again but I have to tell you: My usual inclination if approached to re-offer is to reduce what I'm willing to pay.  Have your people call my people.
All in all you're just another brick in the wall. - Roger Waters
nfs
***** God Member
 
Posts: 537
Joined: 03/14/05 at 13:21:11
Location: The Dry Side of the Wet Coast

Next

Return to Honest and Informed Early Retirement Board

Who is online

Users browsing this forum: Bing [Bot] and 1 guest

cron