PP Buyers Remorse...Some Sound Feedback Needed!

[LNGTERMER]

T

ypically over the long term (i.e. backtested from 1800) it paces buy and hold on a long term return basis, but periodically leads and at other times lags.

A question for Clive or any one who who can answer this, is how you obtain such data that goes back to 1800s? I have access to the free yahoo data that may go back to the 1950s.

Another style you might like to test is what I call SL7.

With AmiBroker it is not difficult to back test if you know a programming language such as C for example, but it does require time for development, debugging, ..etc. I will have to look into this. As I alluded to previously I have stumbled into the PP recently and I was skeptical at first but once I started testing it did test well. I have been developing systems for my own benefit for a while and experience tought me that it's very easy to program own biases into the system. These include time periods and scale, over optimization, and curve fitting among others. What I like about the PP is that the results are collaborated by so many others eliminating my own personal biases or blind spots.

I am also starting to be fascinated by the philosophies behind the each element of the PP, so I am reading to get a better understanding about them. Gold for example and fiat money is a new topic I am exploring. An other is the debt monitory system and how it's tilted toward banks and how the whole thing encourages more debt and growth at any cost just to keep with currency loosing value over time.

The most important thing I got from the PP is risk reduction, I really don't like loosing money I diligently earned, so for now I am happy with basic PP. I might explore the VP at some point, but only after testing while at the same time retaining the principles that gives the PP its strength.

[pumpkin]

Lastly, there is this fallacy of time diversification. This is the idea that stocks get less risky the longer you hold them - this is a lie. They in fact get riskier the longer your are in the game. John Norstad wrote a paper on this very subject:

Craig, thanks for the link to that article!  I've seen this in FireCalc, but now it's explained.

For those of who might not know about FireCalc, it's a neat tool that uses historical data to plot all the historical outcomes of your particular strategy.   In other words, if I have 30 years until my retirement, it will take my inputs and run the strategy from 1914 to 1944.   It plots the outcome.  Then it runs 1915-45 and plots the outcome.  Then each year until 1979-2009 (or so).  This gives you an range of returns for all the historical 30 year periods.  I wish I could figure out how to post a screen shot to show you, but I can't, so I'll try to give you some results to illustrate.




As a starting point, I have a portfolio of $10K that I put in the S&P500 for 30 years.
After 5 years, the portfolio would have been worth between ~$10-28K
After 10 years, between ~$10-42K
After 20 years, between ~$12-114K
After 30 years, between ~$30-189K

So the range of possible outcomes gets really wide as the time invested increases.  

Try it for yourself here: http://www.firecalc.com


[EDIT: Added chart, thanks for the posting instructions foglifter!]

[foglifter]

I wish I could figure out how to post a screen shot to show you, but I can't, so I'll try to give you some results to illustrate.

pumpkin, here's how you can post the image
1. Make a screen shot of the particular window you want to post using Windows snipping tool or via Alt+Print
2. Save it in a file on your computer
3. Go to tinypic.com and upload the file
4. On the page "Share This Image" select and copy into clipboard the string "IMG Code for Forums & Message Boards"
5. Paste it into your forum post body. You are done!

[LNGTERMER]

Thanks for sharing these great recourses, much appreciated. 

[macclary]

http://www.jeremysiegel.com/index.cfm/f ... e/data.cfm

Siegel has fairly long data sets posted (for a price). If he thinks he can charge for them, then maybe that means that it would take some elbow grease to get freely available data massaged into usable form.

@Bonafede: my comments about Bogle and others looking at portfolio components in isolation comes down to this: why no gold? why no long bonds?

Non PP Answer about gold: if you look at gold in isolation it doesn't act like a stock or a bond.

Non PP regarding long bonds: if you look at long bonds in isolation their risk adjusted returns aren't as good as shorter durations.

If the Bogle et al actually looked at whole portfolios over many time periods then they would include significant amounts of gold and long bonds in their policy portfolios just like Browne did. qed

Markowitz (as well as Sharpe and others) are bright guys who did advance "modern portfolio theory". BUT they have probably lost more money for their followers than if they hadn't ever published. This is because of the two tragic errors at the core of MPT.

1) Variance as a (rather sorry) proxy for risk.
2) Optimization framework that looks for maximum return instead of maximum safety.

These assumptions cut through the heart of MPT. Browne rightly rejected both of them.

[craigr]

Looking at assets in isolation is one of the most common mistakes I see in investment literature that discusses portfolio construction and diversification.

[craigr]

This is because of the two tragic errors at the core of MPT.

1) Variance as a (rather sorry) proxy for risk.
2) Optimization framework that looks for maximum return instead of maximum safety.

These assumptions cut through the heart of MPT. Browne rightly rejected both of them.

My wife came up with a great way to describe the issue of holding gold and cash in a portfolio vs. holding all "growth" assets.

Assets like cash and gold are the same thing as ropes and passive protection devices are to rock climbers. Now there are two ways to look at this:

School #1: The "you need growth and protection" school (Permanent Portfolio) says that ropes and chocks can be dead weight at times, but the day you fall is the day you realize just how happy you are that you carried them.

School #2: The "you need growth only" school says that the ropes and protection chocks a climber carries are useless because, after all, they don't help you climb. These things are just weighing you down and rarely are they needed so why bother?


I think School #1 is most realistic.

[Bonafede]

@Bonafede: my comments about Bogle and others looking at portfolio components in isolation comes down to this: why no gold? why no long bonds?

Non PP Answer about gold: if you look at gold in isolation it doesn't act like a stock or a bond.

Non PP regarding long bonds: if you look at long bonds in isolation their risk adjusted returns aren't as good as shorter durations.

If the Bogle et al actually looked at whole portfolios over many time periods then they would include significant amounts of gold and long bonds in their policy portfolios just like Browne did. qed

Markowitz (as well as Sharpe and others) are bright guys who did advance "modern portfolio theory". BUT they have probably lost more money for their followers than if they hadn't ever published. This is because of the two tragic errors at the core of MPT.

1) Variance as a (rather sorry) proxy for risk.
2) Optimization framework that looks for maximum return instead of maximum safety.

These assumptions cut through the heart of MPT. Browne rightly rejected both of them.

Hi Macclary,

Thanks for the post. Great points on the MPT and EMH regarding variance and optimization. I think the theory is overall sound and has provided a whole new way to look at constructing portfolios...but agree that there are some very big challenges to the theories...particularly with risk/return.

Below is actually a great website from a fellow investor / financial planner in Australia that makes some good points about challenges to the MPT and EMH:

http://travismorien.com/invest_FAQ/content/view/220/58/

Regarding Bogle and other's looking at assets in isolation, I think is misleading and in fact incorrect. This contradicts MPT and what these gents advocate. At the heart of MPT is finding assets that behave differently from one another, then holding them together as less risk than individually. Harry Browne took this one step further and devised a portfolio that generates fairly good returns based not on how assets act towards each other, but rather how they act/react to the economy. Key difference.

Also, Bogle wasn't necessarily against Gold. In fact he stated in his book, Bogle on Mutual Funds the following:

"Finally a word about concentrated specialty funds. Some may have a role as part of your mutual fund portfolio. Others will not. In my view, among the more useful types may be gold and energy funds,(seeking to provide a hedge against inflation), utility funds(seeking to provide a higher level of income than equity income funds), and technology and health care funds(enabling an investor to own segments of the US economy that may provide above average long term growth). It is difficult to generalize as to the proportion of your equity portfolio you should allot to these concentrated funds. It might be as much as 25% in the case of utility funds ( if you require more income) or 10% in healthcare or technology funds (if you are willing to incur the additional risk)."

This is from "Bogle on Mutual Funds" Pg 76

Everyone has pros/cons for not including or excluding gold from a portfolio. From the perspective of HB PP, I'm all for it and get the importance of having it in the portfolio.

Thanks again for the post. This is great stuff!

-b

[macclary]

Great points on the MPT and EMH regarding variance and optimization. I think the theory is overall sound...

I wouldn't call MPT sound. It doesn't actually work in practice, especially for volatile asset classes like equities. That is why few people build portfolios solely with MPT. For low vol portfolios MPT may be better than no theory at all. For higher vol portfolios though MPT will eat your money over time because it chases over-valued assets.

Thanks for trying to correct my knowledge of Bogle's understanding of MPT. Here is some evidence that lead me to think that Bogle does not analyze whole portfolios before talking:

1) The Bogle lazy portfolio is just an ad-hoc collection of funds. It represents conventional wisdom not MPT or any other portfolio science:

Vanguard Total Stock Market Index      VTSMX  40%
Vanguard Interm-Term U.S. Treas        VFITX  30%
Vanguard Total Intl Stock Index        VGTSX  10%
Vanguard Inflation-Protected Secs      VIPSX  10%
PIMCO Foreign Bond                      PFUIX  10%

note 1: lack of gold, long bonds, real estate, commodities, etc. that MPT or other methodologies would have included.

note 2: portfolio is weighted arbitrarily not based on volatility, drawdown, global market portfolio, economic climates or growth projections.

2) Bogle regularly recommends that 20 year olds own 80% equities, based on a counter-scientific age in bonds scheme. No one should own 80% equities. This is irresponsible advice that does not take diversification benefits into account.

3) The Bogle quote shared regarding "concentrated specialty funds" shows a mis-understanding of MPT. Bogle refers to diversification into technology and health care funds as an option for higher growth at the beginning of the paragraph, but at the end he forgets what he was saying and adds "...or 10% in healthcare or technology funds (if you are willing to incur the additional risk)." According to MPT diversification can add to returns without adding to risk. (Note that Bogle must be talking about diversification rather than just increasing the beta of the portfolio, since equity beta could be purchased more cheaply using a higher allocation to the Vanguard TSM or S&P 500 funds which he sells.)

Hopefully this analysis of Bogle is useful to some people who wonder why more experts don't embrace the PP  -- in truth such experts may not even understand Markowitz let alone Browne.

[Pkg Man]

Great points on the MPT and EMH regarding variance and optimization. I think the theory is overall sound...

I wouldn't call MPT sound. It doesn't actually work in practice, especially for volatile asset classes like equities. That is why few people build portfolios solely with MPT.

macclary can you cite any academic or other articles that have come to that conclusion?  I am not doubting you at all, I am just curious to see what some of the literature says.

[Bonafede]

Great posts, and thanks for the feedback! Just for kicks, since we're talking about Harry Markowitz, here's a recent interview with him from the Journal of Financial Planning:

http://www.fpajournal.org/CurrentIssue/ ... Questions/

He makes some interesting comments in there about how the portfolios constructed based on MPT performed as expected in the most recent market challenge. Not sure I agree completely, but interesting still.

[craigr]

The major item that appealed to me about Browne's approach is that he had good economic reasoning why his portfolio worked. I think it goes beyond MPT because the assets chosen were specific in their function and purpose in the economy and portfolio. Many portfolio schemes I come across seem more like a random guess of flinging a bunch of assets at the wall and watching what sticks. I see no particular logic or economic reasoning behind the recommendations (especially the very complicated portfolios which are trying to appear sophisticated by using small slices of lots of assets).

Even beyond this, Browne does not pay attention to the idea of asset class correlations with each other - Another common mistake IMO. He is only concerned with how the assets correlate to the economy which is what's really important. It's also the main reason why the portfolio has been so robust.

[Pkg Man]

Perhaps Brett has been reading this forum? 
http://online.wsj.com/article/SB1280001 ... stpop_read

While not exactly espousing the PP strategy, this is about as close as I've seen in a main-stream publication.  Very HB-like IHMO.

[Bonafede]

Perhaps Brett has been reading this forum? 
http://online.wsj.com/article/SB1280001 ... stpop_read

While not exactly espousing the PP strategy, this is about as close as I've seen in a main-stream publication.  Very HB-like IHMO.

That is an excellent article!!! Thanks for sharing that! Also, thanks for everyone else who contributed on this thread. It's been very helpful to me as a new PP adopter. The more I've read the last few days, the more comfortable I am with this approach. Thanks again and very appreciated!!

-b

[macclary]

Hi Pkg Man, I don't have any specific academic articles in mind. The article that Bonafede posted showed some of MPT's short comings. Here is something else I wrote on the topic: http://www.riskcog.com/mean_variance_comparison.jsp

Regarding MPT performing worse with more volatile portfolios. Consider two points: 1) arithmetic returns are bad approximations for geometric returns of volatile assets 2) optimizing for maximum profit doesn't work in walk forward testing or real life.

Recall that MPT is built around average returns and volatility. Average returns here are "arithmetic average returns". Recall also that arithmetic returns predict actual geometric returns more badly with increasing volatility. This is sometimes called "volatility drag". Here is an explanation:

http://tablerockfinancial.com/index.php ... turnid=101

That link gives an example of assets A and B which have respective Arithmetic average returns of 8.0% and 8.5% respectively, but geometric returns of 8.0% and 7.2% respectively. In the MPT model B would be considered higher return with higher volatility compared to A, but in real life B is LOWER return with higher volatility.

Point 2: mean-variance optimization which falls out of MPT, doesn't work very well especially for volatile assets. MPT is supposed to use predicted returns and variance of the assets it optimizes, but perfect prediction is a hard problem so practitioners substitute historic returns. This means that at the top of an equity bull market, mean-variance optimized portfolios are loaded with the "low volatility" and "high return" equity funds. At the bottom of a bear market MVO portfolios emphasize bond funds since the equity funds have been so volatile with little return. 

MVO and other frameworks which try to optimize return for a certain risk level, fall into a pro-cyclical trap which exposes the investor to the most equities right before a crash. Basically buy-high sell-low. This is like accumulating stocks by buying a fixed number of shares instead of a fixed dollar amount - the mathematics just don't work well for the stock market. For non-volatile asset classes buying fixed "shares" or dollars is roughly equivalent, but for volatile assets you have to be much more careful to not buy tops or sell bottoms.

The alternative is to optimize for the safest portfolio for a target return, which yields a contra-cyclical buy-low sell-high investment pattern.

[macclary]

Thanks for the feedback Clive. Maybe I should put some bold claims on the site so that visitors understand the point: "You could spend hundreds of dollars on MVO portfolio optimization software but the output would not be as good as this free website!!"

Based on your suggestion, I changed the site to put all of the years into the selection box for the end year of the optimization. This is how you do IN and OUT of sample testing: select an end year such as 1990, hit the button which will give you IN sample results. Next change the end year to 2009 (for example): the results past 1990 are OUT of sample.

I think I'll start a new thread in "Other Discussions" to see if people are willing to help me improve the site...

[Pkg Man]

Hi Pkg Man, I don't have any specific academic articles in mind. The article that Bonafede posted showed some of MPT's short comings. Here is something else I wrote on the topic: http://www.riskcog.com/mean_variance_comparison.jsp
...

Thanks for the reply macclary.  Here is something I just found that I will be looking into at some point.  Thought that you might be interested as well.

http://www.albany.edu/~faugere/Outstand ... inance.htm

[Reido]

I just came across the permanent portfolio a couple of months ago myself - I'm 28 yrs old and living in the US.

I wanted to share a bit about my personal findings - I've used the Simba Back-Test portfolio checker and decided that at my age the most appropriate strategy for me, personally, is to start a PP with ETF's (GLD, SHY, TLT, VT) and then to enhance return I've started a VP with VNQ (REIT index) accounting for all of it.  The VP is 1/4 of the value of the PP, so basically its like having a 5x20 portfolio.

As such my total portfolio historically would have outperformed stocks, while the greatest loss it incurs is 7.8% in 2008.  This is a significant loss, but one which I can stand as a younger individual - with time I'll cut-down on the REITs and merge into a complete PP only.

The problem with stocks as per the firecalc outcomes is that lets say hypothetically you were about to retire in 2-3 years and you were at the end of a bull market (as would have happened if you wanted to retire in 1999-2000 or 2006-2007), you may have suddenly suffered a huge loss 20-30%.  This lower value may completely destroy your hopes of retirement at the stated time - but only changes your average return over the course of your "accumulation" period by less than 1%.  You could subsequently see a 10-20 yr period wherein you have no more capital appreciation and your money declines significantly more than you expect.


The likielihood of having a crash occur during retirement seems enormous when you think that you may be retired for 20-30+ years

[6 Iron]

The problem with stocks as per the firecalc outcomes is that lets say hypothetically you were about to retire in 2-3 years and you were at the end of a bull market (as would have happened if you wanted to retire in 1999-2000 or 2006-2007), you may have suddenly suffered a huge loss 20-30%.  This lower value may completely destroy your hopes of retirement at the stated time - but only changes your average return over the course of your "accumulation" period by less than 1%.  You could subsequently see a 10-20 yr period wherein you have no more capital appreciation and your money declines significantly more than you expect.

This is what happens when investors confuse risk tolerance with greed. As you note, a skewed portfolio, traditionally to stocks, significantly increases your chances of a big draw down at, or early in, retirement.

[Pkg Man]

...
The problem with stocks as per the firecalc outcomes is that lets say hypothetically you were about to retire in 2-3 years and you were at the end of a bull market (as would have happened if you wanted to retire in 1999-2000 or 2006-2007), you may have suddenly suffered a huge loss 20-30%.  This lower value may completely destroy your hopes of retirement at the stated time - but only changes your average return over the course of your "accumulation" period by less than 1%.  You could subsequently see a 10-20 yr period wherein you have no more capital appreciation and your money declines significantly more than you expect.

The likielihood of having a crash occur during retirement seems enormous when you think that you may be retired for 20-30+ years

An excellent point.  The possibility of such a scenario is what motivated me to find something safer and I feel lucky to have found the PP.

[Reido]

I admit I have only a cursory understanding of this topic, but one other thing I've noticed (and I know books like Supercycles discuss this as well) is that there are long periods of time - 10-15 year periods in which the stock market performs poorly and 10 - 15 year periods where returns are terrific. 

These periods may be underestimated if you break down performance into 20 year periods from 1900-1919 and 1901-1920, 1902-1921... and so on, because there will be amplification of the effect of these years upon your portfolio with time...  The last year before retirement - or even the last year before you move some assets out of the market will have the greatest effect on your portfolio.

If you think slowly putting money into bonds will save you, it certainly may help, of course if stagflation comes along - you'll be SERIOUSLY hit and ANY kind of high inflation will hurt - A LOT.