100% stocks for 30y+: thinking/ challanging/ optimizing

[nugget]

Dear all,
motivated by recent 100% stocks threads and replies like

100% stocks allocation suffers from two flaws

i thought i'd expose my own portfolio for some thought and interesting inputs this forum might have to give me ;)
My dream portfolio would be to own an equal amount of $$ in every liquid stock in the world. As this is rather difficult, i approximate it by the following. note the small, value & em weights:

15% VTI, Vanguard US total stock market
15% VXF, "VTI ex SP500"
15% VBR, Vanguard US Small Value
20% VXUS, Vanguard total world exUS
20% VSS, Vanguard total world exUS small
15% VWO, Vanguard emerging markets

+ cash cushion
+ some social security that i dont want to include here

I did put lots of though into it. But of course, without disputing about it with others i cant optimize it ;)
Any ideas/ comments on how to optimize this?

[caffeine]

I wouldn't invest in emerging markets. The same emerging markets have been emerging markets for over 120 years.

Investing in small cap US & S&P 500 seems redundant with the Total Market index. It would seem better to me if you just did the Total Market.

By investing in the largest US companies, you are already invested in international companies that aren't acting solely in the US. The expense to run those funds would also be cheaper than the ex US pieces.

[TomTX]

I wouldn't invest in emerging markets. The same emerging markets have been emerging markets for over 120 years.

Investing in small cap US & S&P 500 seems redundant with the Total Market index. It would seem better to me if you just did the Total Market.

By investing in the largest US companies, you are already invested in international companies that aren't acting solely in the US. The expense to run those funds would also be cheaper than the ex US pieces.

I believe you inverted the holdings of the second fund. That fund appears to be VTI, but excluding the S&P 500.

Basically reducing the overweighting in large caps that VTI has inherently.

[caffeine]

I wouldn't invest in emerging markets. The same emerging markets have been emerging markets for over 120 years.

Investing in small cap US & S&P 500 seems redundant with the Total Market index. It would seem better to me if you just did the Total Market.

By investing in the largest US companies, you are already invested in international companies that aren't acting solely in the US. The expense to run those funds would also be cheaper than the ex US pieces.

I believe you inverted the holdings of the second fund. That fund appears to be VTI, but excluding the S&P 500.

Basically reducing the overweighting in large caps that VTI has inherently.

You are correct. My mistake.

[SeattleCPA]

Dear all,
motivated by recent 100% stocks threads and replies like

100% stocks allocation suffers from two flaws

i thought i'd expose my own portfolio for some thought and interesting inputs this forum might have to give me ;)
My dream portfolio would be to own an equal amount of $$ in every liquid stock in the world. As this is rather difficult, i approximate it by the following. note the small, value & em weights:

15% VTI, Vanguard US total stock market
15% VXF, "VTI ex SP500"
15% VBR, Vanguard US Small Value
20% VXUS, Vanguard total world exUS
20% VSS, Vanguard total world exUS small
15% VWO, Vanguard emerging markets

+ cash cushion
+ some social security that i dont want to include here

I did put lots of though into it. But of course, without disputing about it with others i cant optimize it ;)
Any ideas/ comments on how to optimize this?

These two thoughts:

1. Do you know your portfolio's standard deviation? And are you okay with that?

2. Do you understand how modern portfolio theory works (which is what i talked about in that blog post you pointed to)?

BTW, if someone answers "yes" to both of the above questions, I think you're good. With a portfolio like yours, I'd personally feel I was bearing too much risk. But if you understand that risk and the practical way you can manage it, you have my blessing (for what that's worth).

Where things get dicey, I think, is when someone doesn't really understand the risks in their portfolio. Without that understanding, they can't really know whether or not what they're doing makes sense.

Two other semi-random thoughts:

1. As I've noted before, my asset allocation for traditional asset classes uses David Swensen's formula: 30% US stocks; 15% each Developed international, REITs, imtermediate treasuries and TIP; 10% emerging market. I've also tried to have meaningful alternative asset investments like direct real estate and small business.

2. In thinking about applying modern portfolio theory to individual investors, I am reminded of what guitarist Robben Ford said about Miles Davis. "People have said a lot of good things about Miles and a lot of bad things. And what you need to remember is that everything was true."

[Telecaster]

Dear all,
motivated by recent 100% stocks threads and replies like

100% stocks allocation suffers from two flaws

i thought i'd expose my own portfolio for some thought and interesting inputs this forum might have to give me ;)
My dream portfolio would be to own an equal amount of $$ in every liquid stock in the world. As this is rather difficult, i approximate it by the following. note the small, value & em weights:

15% VTI, Vanguard US total stock market
15% VXF, "VTI ex SP500"
15% VBR, Vanguard US Small Value
20% VXUS, Vanguard total world exUS
20% VSS, Vanguard total world exUS small
15% VWO, Vanguard emerging markets

+ cash cushion
+ some social security that i dont want to include here

I did put lots of though into it. But of course, without disputing about it with others i cant optimize it ;)
Any ideas/ comments on how to optimize this?

Just spit ballin'....

Foreign stock ETFs tend to have higher fees.  So that will be a drag on whatever outperformance you hope to get.  Speaking of which, what is the rationale behind wanting to own the whole universe of stocks?   

[MustacheAndaHalf]

Just to refute the claim about expensive international funds:

VTI, US stocks, 0.04%
VXUS, international, 0.11%

Over 42 years, 0.07% compounds to 3%.  Meaning $1,000,000 v.s. $1,030,000 assuming international diversification has 0.00% impact on your returns.  Plus, most people have a home country bias and tend to have about 1/3rd international... which means a total impact over 42 years of roughly 1% ($1,000,000 vs $1,010,000).

[harvestbook]

I don't get the logic that "US stocks expose you to the whole world anyway." That's factually false. Many foreign companies, especially smaller ones, have little or no business outside their borders. My thinking is I want to get a piece of the action anytime someone buys manga in Japan, vegemite in Australia, or guinea pig meat in Peru. True, a US-only portfolio might get the job done, but it doesn't seem either optimal or safe to me. I prefer as much diversity as possible, because I doubt the future will look like the past.

[nugget]

Thanks for all the replies, maybe you have even more?
to comment/ reply on some lines above:

Do you know your portfolio's standard deviation? And are you okay with that?
Do you understand how modern portfolio theory works

I'd answer both with yes. I plan to not use any of this money before retirement

what is the rationale behind wanting to own the whole universe of stocks?

The rationale is maximum diversification in the asset class of stocks. I chose stocks as asset class because everything that provides more long term returns is either not suited for private low-effort investors (Venture capital....) or excessively speculative (derivatives, crypto, VIX,...)

[boarder42]

I wouldn't invest in emerging markets. The same emerging markets have been emerging markets for over 120 years.

Investing in small cap US & S&P 500 seems redundant with the Total Market index. It would seem better to me if you just did the Total Market.

By investing in the largest US companies, you are already invested in international companies that aren't acting solely in the US. The expense to run those funds would also be cheaper than the ex US pieces.

VTI is market weighted so its essentially an S&P500 fund due to the size of those companies it holds very little on the small and midcap front - if you want to increase returns and volatility adding some small and midcaps does this where as adding more VTI vs VOO wouldnt really effect it much.

[boarder42]

I don't believe in MPT.  I view MPT is a mathematical religion constructed around the belief that risk = volatility.  Unfortunately, for most folks who are not short term traders, volatility != risk. 

Risk is defined as the likelihood that the purchasing power of an investment after all taxes, fees, and inflation will not be maintained over the holding period.

On a short term time scale used by traders: say a matter of weeks, months, or up to a year or two, volatility is a good approximation of risk, because swings in the asset price will create a very real risk of the asset failing to maintain its purchasing power over the short time frame.

However if someone has 10 or 20 or 30 years to invest, volatility is less meaningful.  What is far more risky is that inflation, taxes, and fees will consume too much of the real value of an asset, or of an entire portfolio leaving the investor with less purchasing power at the end of the 10, 20, or 30 year holding period.

Take for instance a current 10 year bond yielding under 3%.  I would say this is an incredibly risky investment to buy and hold for 10 years.  The risk of loosing real, net purchasing power is virtually 100%.  Whereas a diversified group of stocks will have a much higher likelihood of maintaining (and most likely increasing) purchasing power.  I think the differences get even more stark with 20 year bond vs. stocks, or 30 year bond vs. stocks.

So why do we have MPT?  I think we have it because a bunch of not-actually-rich academics were able to produce some beautiful equations and charts and graphs when they substituted (past) volatility for true risk and looked at historical data.  Greek letters could be invoked, people could have equations and ratios named after them.  Some could even win Nobel prizes and thereby become anointed.  It has become an academic's nirvana.  When things get wonky, they show that the math still holds if the event is considered exceptionally rare....  The joke goes, "It's more convenient to add a sigma, than to change a formula."

I would argue in fact that the entire financial crisis of '08/09 was exacerbated by the use of MPT to the exclusion of common sense. One idea comes to mind which caused people to ignore enormous risk, namely  VAR.

VAR stands for Value at Risk, and it must be computed by banks regularly to determine how much "risk" is in a portfolio. But it is a statistical method of computing risk based on passed volatility.  So for example a collection of shit-loans sub-prime loans were rated investment grade and have very low statistical risk, whereas a junk-bond that has already dropped like a rock and was rated below investment grade but was trading below a conservative estimate of available assets of the firm would be considered "risky."  The result is that the purchase of the statistically less risky assets were allowed to be leveraged whereas the purchase of the statistically more risky (but actually less real risk because it is backed by real assets) would likely have to be hedged against.  Some risk management system.

this is basically the way my brain works when we talk about risk and why people who are thinking they are being conservative - in my opinion based on 99% of what you just said are actually creating more risk than they think. 

[steveo]

However if someone has 10 or 20 or 30 years to invest, volatility is less meaningful.  What is far more risky is that inflation, taxes, and fees will consume too much of the real value of an asset, or of an entire portfolio leaving the investor with less purchasing power at the end of the 10, 20, or 30 year holding period.

Take for instance a current 10 year bond yielding under 3%.  I would say this is an incredibly risky investment to buy and hold for 10 years.  The risk of loosing real, net purchasing power is virtually 100%.  Whereas a diversified group of stocks will have a much higher likelihood of maintaining (and most likely increasing) purchasing power.  I think the differences get even more stark with 20 year bond vs. stocks, or 30 year bond vs. stocks.

I think you are missing a key point here. If the stock market tanks (which it will) and you withdraw too many stocks at that point you can severely stuff up your net worth. You can stuff it up so bad that you fail at retirement.

Yes the no 1 source of retirement failure will be inflation and you need a high stock allocation to handle that scenario however you also need some safety net to cater for a tanking stock market and therefore depleting your stocks at a time when it critically impacts your net worth.

You can create that safety net by bonds or a bigger stache (this is not ideal) or by going back to work (again not ideal). The question is how do you create that safety net or are you just praying that you don't get caught out by a declining stock market.

[SeattleCPA]

I don't believe in MPT.  I view MPT is a mathematical religion constructed around the belief that risk = volatility.  Unfortunately, for most folks who are not short term traders, volatility != risk. 

Risk is defined as the likelihood that the purchasing power of an investment after all taxes, fees, and inflation will not be maintained over the holding period.

On a short term time scale used by traders: say a matter of weeks, months, or up to a year or two, volatility is a good approximation of risk, because swings in the asset price will create a very real risk of the asset failing to maintain its purchasing power over the short time frame.

However if someone has 10 or 20 or 30 years to invest, volatility is less meaningful.  What is far more risky is that inflation, taxes, and fees will consume too much of the real value of an asset, or of an entire portfolio leaving the investor with less purchasing power at the end of the 10, 20, or 30 year holding period.

Take for instance a current 10 year bond yielding under 3%.  I would say this is an incredibly risky investment to buy and hold for 10 years.  The risk of loosing real, net purchasing power is virtually 100%.  Whereas a diversified group of stocks will have a much higher likelihood of maintaining (and most likely increasing) purchasing power.  I think the differences get even more stark with 20 year bond vs. stocks, or 30 year bond vs. stocks.

So why do we have MPT?  I think we have it because a bunch of not-actually-rich academics were able to produce some beautiful equations and charts and graphs when they substituted (past) volatility for true risk and looked at historical data.  Greek letters could be invoked, people could have equations and ratios named after them.  Some could even win Nobel prizes and thereby become anointed.  It has become an academic's nirvana.  When things get wonky, they show that the math still holds if the event is considered exceptionally rare....  The joke goes, "It's more convenient to add a sigma, than to change a formula."

I would argue in fact that the entire financial crisis of '08/09 was exacerbated by the use of MPT to the exclusion of common sense. One idea comes to mind which caused people to ignore enormous risk, namely  VAR.

VAR stands for Value at Risk, and it must be computed by banks regularly to determine how much "risk" is in a portfolio. But it is a statistical method of computing risk based on passed volatility.  So for example a collection of shit-loans sub-prime loans were rated investment grade and have very low statistical risk, whereas a junk-bond that has already dropped like a rock and was rated below investment grade but was trading below a conservative estimate of available assets of the firm would be considered "risky."  The result is that the purchase of the statistically less risky assets were allowed to be leveraged whereas the purchase of the statistically more risky (but actually less real risk because it is backed by real assets) would likely have to be hedged against.  Some risk management system.

Maybe we mean different things by MPT... But practically what MPT means to me are things like this: If you take two investments that generate 9.5% annually but aren't perfectly correlated, you can blend them and earn more than 9.5% and yet still have no more volatility. (This was something I illustrated in my blog post which OP pointed to.)

Another thing that MPT means to me (also illustrated in the referenced post): You might be able to generate 9.5% return with a high volatility portfolio or generate a 9.5% return with a not-so-high volatility portfolio. Why not take the lower volatility choice?

BTW, for everyone who gets the preceding and still thinks MPT is problematic? Hey, I probably agree with you more than you might guess.

[boarder42]

However if someone has 10 or 20 or 30 years to invest, volatility is less meaningful.  What is far more risky is that inflation, taxes, and fees will consume too much of the real value of an asset, or of an entire portfolio leaving the investor with less purchasing power at the end of the 10, 20, or 30 year holding period.

Take for instance a current 10 year bond yielding under 3%.  I would say this is an incredibly risky investment to buy and hold for 10 years.  The risk of loosing real, net purchasing power is virtually 100%.  Whereas a diversified group of stocks will have a much higher likelihood of maintaining (and most likely increasing) purchasing power.  I think the differences get even more stark with 20 year bond vs. stocks, or 30 year bond vs. stocks.

I think you are missing a key point here. If the stock market tanks (which it will) and you withdraw too many stocks at that point you can severely stuff up your net worth. You can stuff it up so bad that you fail at retirement.

Yes the no 1 source of retirement failure will be inflation and you need a high stock allocation to handle that scenario however you also need some safety net to cater for a tanking stock market and therefore depleting your stocks at a time when it critically impacts your net worth.

You can create that safety net by bonds or a bigger stache (this is not ideal) or by going back to work (again not ideal). The question is how do you create that safety net or are you just praying that you don't get caught out by a declining stock market.

Declining stock markets aren't the end of retirement and can be planned for to increase the safety historically without adding more bonds.  There are many tools that work for this including variable withdrawal methods. Very very rarely has a decline stock market been bad enough to kill a retirement. So there isn't much praying to do. Flexibility is King not creating a conservative AA that is more likely to end up with lower money later in life which adding bonds inherently does.

[SeattleCPA]

I don't believe in MPT.  I view MPT is a mathematical religion constructed around the belief that risk = volatility.  Unfortunately, for most folks who are not short term traders, volatility != risk. 

Risk is defined as the likelihood that the purchasing power of an investment after all taxes, fees, and inflation will not be maintained over the holding period.

On a short term time scale used by traders: say a matter of weeks, months, or up to a year or two, volatility is a good approximation of risk, because swings in the asset price will create a very real risk of the asset failing to maintain its purchasing power over the short time frame.

However if someone has 10 or 20 or 30 years to invest, volatility is less meaningful.  What is far more risky is that inflation, taxes, and fees will consume too much of the real value of an asset, or of an entire portfolio leaving the investor with less purchasing power at the end of the 10, 20, or 30 year holding period.

Take for instance a current 10 year bond yielding under 3%.  I would say this is an incredibly risky investment to buy and hold for 10 years.  The risk of loosing real, net purchasing power is virtually 100%.  Whereas a diversified group of stocks will have a much higher likelihood of maintaining (and most likely increasing) purchasing power.  I think the differences get even more stark with 20 year bond vs. stocks, or 30 year bond vs. stocks.

So why do we have MPT?  I think we have it because a bunch of not-actually-rich academics were able to produce some beautiful equations and charts and graphs when they substituted (past) volatility for true risk and looked at historical data.  Greek letters could be invoked, people could have equations and ratios named after them.  Some could even win Nobel prizes and thereby become anointed.  It has become an academic's nirvana.  When things get wonky, they show that the math still holds if the event is considered exceptionally rare....  The joke goes, "It's more convenient to add a sigma, than to change a formula."

I would argue in fact that the entire financial crisis of '08/09 was exacerbated by the use of MPT to the exclusion of common sense. One idea comes to mind which caused people to ignore enormous risk, namely  VAR.

VAR stands for Value at Risk, and it must be computed by banks regularly to determine how much "risk" is in a portfolio. But it is a statistical method of computing risk based on passed volatility.  So for example a collection of shit-loans sub-prime loans were rated investment grade and have very low statistical risk, whereas a junk-bond that has already dropped like a rock and was rated below investment grade but was trading below a conservative estimate of available assets of the firm would be considered "risky."  The result is that the purchase of the statistically less risky assets were allowed to be leveraged whereas the purchase of the statistically more risky (but actually less real risk because it is backed by real assets) would likely have to be hedged against.  Some risk management system.

Maybe we mean different things by MPT... But practically what MPT means to me are things like this: If you take two investments that generate 9.5% annually but aren't perfectly correlated, you can blend them and earn more than 9.5% and yet still have no more volatility. (This was something I illustrated in my blog post which OP pointed to.)

Another thing that MPT means to me (also illustrated in the referenced post): You might be able to generate 9.5% return with a high volatility portfolio or generate a 9.5% return with a not-so-high volatility portfolio. Why not take the lower volatility choice?

BTW, for everyone who gets the preceding and still thinks MPT is problematic? Hey, I probably agree with you more than you might guess.

I don't think its possible to know that two investments will have the same future returns, say 9.5% but that one will have lower volatility a priori.  Nevertheless, assuming arguendo it is...

Is lower volatility really better for someone who has a negative withdrawal rate (i.e. is in the accumulation phase)?  By selecting lower volatility, they are virtually guaranteeing that they will have to pay more in the future for future investments over investing in the higher volatility portfolio albeit same return portfolio.  There will be less opportunity to buy when things go on sale, so to speak.  I haven't run the numbers but I would hazard a guess that in most runs, in a negative withdrawal rate situation its better to have higher volatility than lower volatility for a given return like say 9.5% (assuming the investor can stick to it) since at the end of the investment period the person who invested into the higher volatility assets will have bought more shares at a lower average basis, and therefore higher gains, than the person who invested in the lower volatility portfolio.

In my free downloadable e-book, Thirteen Word Retirement Plan, I talk about sequence of returns risk using an example like this.

Say you have a $1,000,000 investment portfolio that generates exactly (say) 6% rate of return so zero variability. Let's assume zero inflation to keep the math simple. In this case, you can draw $60,000 annually and at the end of retirement you still have $1,000,000.

But say you have an investment that averages 6% but annual returns actually equal either 2% or or 10%. Here the sequence of the returns matters, obviously. If you happen to get a string of 2% returns in the beginning, you run out of money if you draw $60,000. If you happen to get a string of 10% returns in the beginning, you never run out of money if you draw $60,000.

Clearly volatility matters to withdrawal rates.

BTW, if you are accumulating and so don't draw anything--say you invest $1,000,000 for ten years and that you're going to retire on whatever that initial investment grows at the end of the ten years--the sequence of returns don't matter. Earning 2% for five years and then 10% for five years gets you to the same future value as earning 10% for five years and ten 2% for five years. (That future value equals $1,778,133.)

Either of these options also gets you to almost the same future value as earning exactly 6% per year for ten years: $1,790,848. (That small shortfall being the volatility drag people talk about.)

For everybody who already understands this, apologies for stating what is obvious to you. But my strong hunch is many people haven't yet internalized the above math. And until they do, I worry they don't understand enough about the risks baked into their portfolios.

[GOFU]

1. Do you know your portfolio's standard deviation? And are you okay with that?

Can you please expound on this idea of standard deviation and how it plays in to real world planning for asset allocation, particularly  for someone trying to decide if he could retire with his current asset allocation?

I think I understand that higher SD means more volatility, which if you are withdrawing from the portfolio could expose you to sequence of returns risk. I know everyone's tolerances vary, but I guess I am looking for some kind of guideline or generally accepted definitions for what is a high or low standard deviation for a portfolio.

Thank you.

[privatefarmer]

I don't get the logic that "US stocks expose you to the whole world anyway." That's factually false. Many foreign companies, especially smaller ones, have little or no business outside their borders. My thinking is I want to get a piece of the action anytime someone buys manga in Japan, vegemite in Australia, or guinea pig meat in Peru. True, a US-only portfolio might get the job done, but it doesn't seem either optimal or safe to me. I prefer as much diversity as possible, because I doubt the future will look like the past.

Bingo. DFAs emerging market fund has had a 0.6 correlation factor with US market over the last 20 years. Still strongly correlated but it has offered some diversification benefit. I don't hold international developed markets, only emerging markets and US market.

[Indexer]

Going to second SeattleCPA. The reason you want bonds is for sequence of return risk in early retirement. If you're 20 years from retirement you can be 100% stocks, and given decent returns you can be 100% stocks again 20 years into retirement. However, right before and especially right after you retire you want some in bonds.

We have all heard of the 4% SWR. The trinity study that came up with the 4% SWR assumed a 60% stock and 40% bond portfolio. Why 60/40?

Let's compare 100% stocks to 60/40.

100% stocks, average returns 10%(going back to the 1920s). 40-50% drops can happen. If you were 100% stocks and your portfolio dropped 50% in a crash, your 4% withdrawal rate turns into an 8% withdrawal rate. As long as the market recovers quick you'll be fine. If it stays low for awhile... that 8% WR will quickly turn into a 9%, 10%, 12%, 15% WR. That becomes unsustainable even if the market recovers and FIRE is over. [Math: 40k off 1 mill is 4%. 40k off 500k is 8%.]

60/40, average returns 8.8%. The worst year for a 60/40 was a 27% drop. If you were drawing 4% before the crash your WR turns to 5.5% after the crash. That's still a low WR rate and it gives your portfolio plenty of time to recover. Plus you would be selling from your bonds, not the stocks, during a down market. A 60/40 also tends to recover much quicker since it didn't drop as much to begin with. Let's talk about the logic of a 60/40 portfolio. With rare exceptions, most periods of negative stock returns last less than 10 years. Knowing that let's say you put 10 years worth of expenses in bonds to weather the next crash, whenever it was. If you are using the 4% rule then you would have 25Xexpenses at retirement. 10Xexpenses in bonds leaves 15Xexpenses in stocks, which equates to a 60% stock and 40% bond portfolio.  IMO 60/40 is the most aggressive you should be on the day you retire if you are withdrawing 4%. [Math: 40k off 1 mill is 4%. 40k off 730k is 5.5%]

Let's fast forward a bit. Let's say you FIRE and your portfolio is growing faster than you need it to. One day you realize you are at a 3% WR. Now sequence of return risk is less of a concern, so you can get more aggressive.


My personal plan:  at FIRE have 10Xexpenses in bonds. During down markets spend that down. As my portfolio, hopefully, grows over time that bond portion will represent less and less as a percentage of my portfolio.

[frugal_c]

Just a few thoughts.   

You are very underweight S&P 500 companies.  The market seems broadly expensive, it's not isolated to S&P500 stocks being overvalued, so I wouldn't underweight them.   

I would go lower on VWO, 10-15% max.  I have owned VWO through a crash and that one more than any other concerned me.  These companies exist in environments where shareholder rights aren't always respected.  I have no doubt that the companies will succeed but I am not sure the shareholders will succeed.  You also already have some emerging exposure with VXUS and VSS.

I would be tempted to do VB over VBR.  There is no right or wrong here but you already have an advantage the way I see it with small-caps.  It just seems that  VB covers a broader range of scenarios.

I could suggest you look at VO (mid-caps).   Someone wrote up a good argument on them awhile back.  Essentially they get much of the superior returns of small-caps but in the event of a financial catastrophe have larger buffers and are more likely to survive.  Personally I lean more towards VO than VB.

All of that being said I am very aware that this is somewhat of a guessing game.  What you have is very reasonable and I am sure you will do well.

[SeattleCPA]

For instance, why ignore inflation?  I think that this is one of the biggest sources of risk in a long-term held portfolio.

I agree inflation is a giant risk. I ignored it in my example (converting everything to real returns and constant-sized dollars) only because I believe working with nominal returns, inflation rates and dollars of different size makes the math even harder for casual readers to understand.

Also, the investment choices you provide are not actually available as far as I know.

The examples I provided in the blog post referenced in the first message in this thread used actual data available at time of blog post.

At that point, the longest history Portfolio Visualizer would model showed nominal return from US Stocks equal to 9.5%.

The longest history PV would model for REITs also showed 9.5%.

And so here's an example of the mathematical insight I think many people miss: If you had blended these two asset classes 50-50, you didn't get 9.5%. Rather you got 10%. And with the same volatility as investing in 100% US stocks.

BTW, anyone who already gets this? Great. I am not worried about your risk awareness. But I wonder... does everybody get why this happens? My hunch is, no, probably not. (Note to these readers: The reason is the rebalancing bonus available because these two asset classes aren't perfectly correlated.)

Also another clarification: I am not saying someone should go 50% US stocks or 50% REITs. I was only using that simple example to illustrate.

You haven't actually addressed the accumulation phase situation that I posed.  In an accumulation phase a person usually starts with almost nothing and has a negative withdrawal rate for 10 or more years.  You substitute a "start with a million dollars" hypo for this.  Not the same thing at all.

I think during accumulation that volatility doesn't matter to returns unless it changes investor behavior. Then it does matter.

E.g. in that blog post referenced, I also pointed out that one could invest in US stocks over the longest period that Portfolio Visualizer would model when I wrote article and you got 9.5% with a 15% standard deviation.

You could have also gotten 9.5% with 35% US stocks, 35% REITs and 30% long treasuries... and an 11% standard deviation. Is that lower standard deviation better? I think it might be if less bouncing about means someone more comfortably stays the course.

Again, though, if someone understands this math and their own personality and says, "Heck makes no difference to me..." that's fine.

My point is, I think lots of people might be surprised by this result too... that an investor might with a 100% stocks be bearing more volatility than they need too.

[TomTX]

You could have also gotten 9.5% with 35% US stocks, 35% REITs and 30% long treasuries... and an 11% standard deviation. Is that lower standard deviation better? I think it might be if less bouncing about means someone more comfortably stays the course.

I am leery of leaning too heavily on long treasuries as an example. Their value to an investor was pumped up for the last ~35 years as interest rates dropped. That trend has recently reversed -  rates are going up. Even if it doesn't - there isn't much room for them to drop any longer.

[steveo]

In my free downloadable e-book, Thirteen Word Retirement Plan, I talk about sequence of returns risk using an example like this.

Say you have a $1,000,000 investment portfolio that generates exactly (say) 6% rate of return so zero variability. Let's assume zero inflation to keep the math simple. In this case, you can draw $60,000 annually and at the end of retirement you still have $1,000,000.

But say you have an investment that averages 6% but annual returns actually equal either 2% or or 10%. Here the sequence of the returns matters, obviously. If you happen to get a string of 2% returns in the beginning, you run out of money if you draw $60,000. If you happen to get a string of 10% returns in the beginning, you never run out of money if you draw $60,000.

Clearly volatility matters to withdrawal rates.

BTW, if you are accumulating and so don't draw anything--say you invest $1,000,000 for ten years and that you're going to retire on whatever that initial investment grows at the end of the ten years--the sequence of returns don't matter. Earning 2% for five years and then 10% for five years gets you to the same future value as earning 10% for five years and ten 2% for five years. (That future value equals $1,778,133.)

Either of these options also gets you to almost the same future value as earning exactly 6% per year for ten years: $1,790,848. (That small shortfall being the volatility drag people talk about.)

For everybody who already understands this, apologies for stating what is obvious to you. But my strong hunch is many people haven't yet internalized the above math. And until they do, I worry they don't understand enough about the risks baked into their portfolios.

I likewise apologize to anyone who finds my arguments obvious. I, too, think that people bear more risk in their portfolios than they realize.  However, I think this risk comes from having too conservative of a portfolio too soon in the accumulation phase.

I don’t have a book.  But I think that JLCollins beautifully describes the situation in the following post:

http://jlcollinsnh.com/2014/06/10/stocks-part-xxiii-selecting-your-asset-allocation/


I'd just like to note @SeattleCPA that I think we agree on more than we disagree.  I concur in the conclusion that once someone is near or has reached the consumption phase that it is a good idea to include some less volatile assets in a portfolio.  As I have said, I personally think that an 80 stock /20 bond portfolio is ideal at this point.  However, I do not agree with analysis that volatility can be categorically substituted for risk in an investment portfolio because I believe that volatility is only a reasonable approximation of risk over short holding periods.  My analysis begins with the expected holding period rather than on the the fact that the price line is more or less squiggly.  I think that for long holding periods, volatility is a bad approximation of risk.  A better gauge of risk over a long time frame is the expected future returns and whether these reasonably stand a chance of preserving purchasing power after all taxes, fees, and inflation are taken into account.  So therefore, a person who is decades from retirement (and therefore decades from the consumption phase) and is in the accumulation phase and has a negative withdrawal rate should be careful not to accumulate too much lower volatility assets that risk loosing purchasing power over the holding period and should seek the highest rate of return.
 
I take issue with your examples, too.  They assume facts and ignore issues in order to support the math.

For instance, why ignore inflation?  I think that this is one of the biggest sources of risk in a long-term held portfolio. Although it might make sense to ignore for a short term holding period, I think its silly to ignore it for a long term holding period like an entire retirement.  This is, in a nut shell, is my criticism of MPT, it ignores to many real world factors and issues and makes too many assumptions in order to "make the math simple." 

Also, the investment choices you provide are not actually available as far as I know.  So I think they make a bad practical examples.  Of course if one is presented hypothetical investment examples, loaded with assumptions, and which ignore key aspects of real risk such as inflation it is easy to funnel people into selecting the lower volatility choice.  But this does not prove that volatility = risk.  And to the extent it implies that volatility is associated with risk, it does not prove that it is categorically a substitute for risk in all situations: particularly in long-term portfolios with a negative withdrawal rate.  I think at most it proves that it can be an approximation of risk for certain shorter holding periods, which is what the first few years of a withdrawal phase are.

You haven't actually addressed the accumulation phase situation that I posed.  In an accumulation phase a person usually starts with almost nothing and has a negative withdrawal rate for 10 or more years.  You substitute a "start with a million dollars" hypo for this.  Not the same thing at all. 

I think that once someone has internalized the math of MPT, it can be hard to see that some of the premises and assumptions might not actually be that sound.   

As the saying goes--

In theory, theory and practice should be the same.
But, in practice they are different.

I agree with your comments here.

The most likely risk to everyone's portfolio is inflation and early in your accumulation phase I do believe that it makes more sense to invest in stocks. I still think though that a portfolio that is about to go into the drawdown phase is much better off having alternative investments like bonds making up a significant (20% or greater) amount of the portfolio.

The only point I'd make is that how much difference does it really make to have 20% of your investments in bonds if that is your longer term goal. Sometimes it might make a difference but other times maybe it is okay.

I'm not sure as well that this is a big criticism of MPT. I think it's just adjusting MPT to cater for a potentially longer term accumulation phase.

[frugal_c]

Have to admit that the returns of the mixed stock / treasury portfolios is impressive. 

I was thinking about the effect of rising rates also.  I ran some numbers and I thought I would share them.

At the beginning of 1972 the 10 yr treasury had a yield of 5.9%.   A decade later the 10 year had gone up to 14%.   During this time the value of a intermediate bond portfolio with a starting $10k value went up to $16.8k.  A US total market portfolio was up to around $20k, a 50 stocks/50 bond portfolio went up to $19.3k.  The portfolios are re-balanced annually.  You can see that the mixed portfolio actually held up quite well.   Unfortunately, I don't have the data for long term treasuries.

[chadat23]

Honest question, but if you don't trust the future to look like the past so you don't trust MPT, then how can you trust that any of this math is relevant given that it's all based on historical data? Is it just that there's more historical data for the total stock market so it's more trustworthy?

Edit: and if this sounds antagonistic it's not meant to be; I think I understand the arguments for and against MPT but that one thing seemed like an inconsistency in the argument against so if I'm simply not fully getting it then I'd love to know where I'm going wrong.

[SeattleCPA]

1. Do you know your portfolio's standard deviation? And are you okay with that?

Can you please expound on this idea of standard deviation and how it plays in to real world planning for asset allocation, particularly  for someone trying to decide if he could retire with his current asset allocation?

I think I understand that higher SD means more volatility, which if you are withdrawing from the portfolio could expose you to sequence of returns risk. I know everyone's tolerances vary, but I guess I am looking for some kind of guideline or generally accepted definitions for what is a high or low standard deviation for a portfolio.

Thank you.

Gofu, I'm not sure if you're asking about what a standard deviation is, what a typical standard deviation is, or how to use a portfolio's standard deviation in an actionable way. Probably some reader wants to know the answer to all three questions, so I'll try to provide a useful answer to all three.

What is Standard Deviation?

It measures the difference or deviation of the return you actually get from the average return. And you can almost think about a standard deviation being the "average deviation" or "average difference" from the average return. E.g., if you buy a investment that generates, always, a 6% return, you always get the average. In this case, the average deviation equals zero. The standard deviation equals zero.

If you invest in an investment that returns 2% half the time and 10% half the time, the average still equals 6%. But on average, a return is 4% more or less than 6%. In this case, the average deviation equals 4%. And the standard deviation equals 4%.

Before someone goes crazy with all this talk about average deviations (which you can measure), let me say that a standard deviation isn't the same thing as an average deviation--though it's close. What a standard deviation calculates (deep breath) is the square root of the average squared difference from the average. This approach means that bigger deviations count more heavily.

For example. if you invest in an investment that returns 0% a quarter of the time, 4% a quarter of the time, 8% a quarter of the time, and then 12% a quarter of the time, the average return still equals 6% and the average deviation still equals 4%, but the standard deviation now equals 4.4721%. What's happened is the standard deviation in effect more heavily weights those values farther from the average.

What are typical standard deviations?

Portfolio Visualizer supplies annualized standard deviations here: https://www.portfoliovisualizer.com/asset-class-correlations

But these values bounce around. BTW, finance professor and writer Jeremy Siegel has calculated that long run stock returns equal 9% and that long run standard deviations of stocks equal 18%.

How does someone use a standard deviation?

We're all probably pretty good at calculating a future value using a rate of return. And you don't get as easy a way to fold in a standard deviation to your calculations.

But, first, I'd think you want to understand that those standard deviations show you (sort of) the average amount the average return will deviate from the percentage you plug into your financial calculator or spreadsheet. I.e., you and I don't want to think "9%" the average, we want to think "well 9% on average but easily anywhere between -9% and +27%."

If you want to get quantitative, you can also build spreadsheets that use your portfolio's expected average return and your portfolio's expected standard deviation to plot a variety of outcomes. Here's a blog post I did that walks someone through steps to building a simple monte carlo simulation workbook in Excel.

https://evergreensmallbusiness.com/stock-market-monte-carlo-simulation-spreadsheet/

[SeattleCPA]

Honest question, but if you don't trust the future to look like the past so you don't trust MPT, then how can you trust that any of this math is relevant given that it's all based on historical data? Is it just that there's more historical data for the total stock market so it's more trustworthy?

Edit: and if this sounds antagonistic it's not meant to be; I think I understand the arguments for and against MPT but that one thing seemed like an inconsistency in the argument against so if I'm simply not fully getting it then I'd love to know where I'm going wrong.

I think above is a really important point.

One can encounter people criticizing MPT-style thinking because past returns, standard deviations and asset class correlations don't let us accurately predict future returns, standard deviations and asset class correlations.

But we also are all (to varying degrees of course) expecting equities to generate higher returns, expecting bonds to generate lower standard deviations, expecting different classes to be something other than perfectly correlated. And these assumptions should mean one should believe that one can improve either a portfolio's return or a portfolio's standard deviation by employing MPT-style thinking.

[steveo]

Honest question, but if you don't trust the future to look like the past so you don't trust MPT, then how can you trust that any of this math is relevant given that it's all based on historical data? Is it just that there's more historical data for the total stock market so it's more trustworthy?

Edit: and if this sounds antagonistic it's not meant to be; I think I understand the arguments for and against MPT but that one thing seemed like an inconsistency in the argument against so if I'm simply not fully getting it then I'd love to know where I'm going wrong.

I think above is a really important point.

One can encounter people criticizing MPT-style thinking because past returns, standard deviations and asset class correlations don't let us accurately predict future returns, standard deviations and asset class correlations.

But we also are all (to varying degrees of course) expecting equities to generate higher returns, expecting bonds to generate lower standard deviations, expecting different classes to be something other than perfectly correlated. And these assumptions should mean one should believe that one can improve either a portfolio's return or a portfolio's standard deviation by employing MPT-style thinking.

Personally I have taken the following points away from MPT:-

1. Indexing works great - you get low fees and diversification.
2. Stocks will tend to outperform all other asset classes over the longer term.
3. Stocks are the most volatile asset class.
4. Diversification across asset classes can decrease your portfolio risk and potentially increase your returns.

When it comes to FIRE I think another key point is that we have to manage the drawdown phase just as much as we manage our savings and portfolio.

Some key points to me in relation to the drawdown phase are:-

1. Sequence of returns risk early in your retirement are the most likely reason your retirement fails assuming you get to a reasonable portfolio size and judge your expenses accurately.
2. A high percentage of bonds in your portfolio is the best way to ensure that your portfolio lasts whilst maintaining a consistent income in the drawdown phase. The high percentage could easily be as high as 50% bonds. I intend to have 30% bonds but this is on the low side based on the analysis that I have read.

[GOFU]

@SeattleCPA  Thanks for the patient and detailed post.

Assume stocks have average annual returns of 9% and a SD of 18%, and bonds have average annual return of 5% with a SD of 6%.
(I'm using the Jeremy Siegel numbers for stocks and making up the bond numbers.)

On a 50/50 portfolio you can expect average annual returns of 7% but it could easily fall between -5% and 19% in any given year. Is this correct to just take a mean of those return and SD values, or is there another concept I am leaving out, covariance or something?

Nobody will complain if the returns deviate to the high side. That is not a "risk." The real risk in maintaining 100% stocks instead of 50/50 is that with 100% you could easily have a down year as low as -9% instead of only -5%.

If all of that is correct, I don't see the grand down side risk to maintaining 100% stocks as compared with 50/50. Four points difference on the extreme down side? Not insignificant and not sustainable, and nobody wants it, but what is the probability really of that occurring? And this compared to an 8 point potential difference on the up side extreme (27% vs. 19%).

If the above is correct arithmetically, it points me to an ideal risk/reward efficiency of something like 75/25 for my rudimentary and simple example. Is this conceptually correct?

Feel free to show me how I am all wet.

[nugget]

Assume stocks have average annual returns of 9% and a SD of 18%, and bonds have average annual return of 5% with a SD of 6%.

I am situated in switzerland, where goverment bonds yield <0%, and corporate bond ETFs up to 1%. once the interest rates rise, these Funds will tank further. You see why i am not very motivated to increase my domestic bond allocation :D

[boarder42]

I think going a bit more conservative once in consumption makes sense, currently we're planning for 90/10 but going conservative prior to consumption i think is flawed.  If the market drops 37% the year before you FIRE you likely arent going to FIRE whether you had 80/20 or 100% stocks AA.  the likelihood of the drop the year prior to FIRE is greatly out weighed by the chance for increase allowing you to FIRE earlier or with a larger nest egg both of which are more conservative than going to bonds too early.

[SeattleCPA]

@SeattleCPA  Thanks for the patient and detailed post.

Assume stocks have average annual returns of 9% and a SD of 18%, and bonds have average annual return of 5% with a SD of 6%.
(I'm using the Jeremy Siegel numbers for stocks and making up the bond numbers.)

On a 50/50 portfolio you can expect average annual returns of 7% but it could easily fall between -5% and 19% in any given year. Is this correct to just take a mean of those return and SD values, or is there another concept I am leaving out, covariance or something?

Nobody will complain if the returns deviate to the high side. That is not a "risk." The real risk in maintaining 100% stocks instead of 50/50 is that with 100% you could easily have a down year as low as -9% instead of only -5%.

If all of that is correct, I don't see the grand down side risk to maintaining 100% stocks as compared with 50/50. Four points difference on the extreme down side? Not insignificant and not sustainable, and nobody wants it, but what is the probability really of that occurring? And this compared to an 8 point potential difference on the up side extreme (27% vs. 19%).

If the above is correct arithmetically, it points me to an ideal risk/reward efficiency of something like 75/25 for my rudimentary and simple example. Is this conceptually correct?

Feel free to show me how I am all wet.

GoFu, I think you get it. You are not all wet. (FWIW, in my traditional asset class investing, I'm 30% US treasuries and 70% equities.)

The one thing I'd add, though, is that one does want to understand the risk.

A while back I did a blog post. Myth of the Long-run Stock Market Return Chart that pointed to the chart below saying people often use charts like this to say that "over the long run, equities don't have much risk..."



People look at posts like this and sometimes erroneously think, hey, over the long run things are sure to work out fine. (Essentially what people are doing is assuming the tail of the funnel chart narrows so if you can just hold on long enough, you can be assured of a average-like return.)

But a better chart for highlighting the range of outcomes is the one show below (from the same blog post):



Note that this line chart tells you it could take (with 75% stocks and 25 bonds) somewhere roughly 15 years and 40 years to get to a $1,000,000 if you save $10K annually.

Furthermore, while the average nestegg after 30-35 years is probably right around $1M, historically someone could have ended up with about $400K (worst case) or as much as roughly $1.5M best case).

[boarder42]

those graphs would be better if it showed the range of outcomes - i know its worst and best - but i think more outcomes skew towards the center and above than towards the lower or worst outcome situation.

[chadat23]

However, those who don't follow MPT but take a more value oriented approach will look to the valuation asset as they are currently buying it as well as using history as a guide.  For instance, right now long term treasury bonds are paying about 3% $30 annually for each $1000 invested.  And as it is fixed in come, the $30 will not grow.   I consider this an insanely risky investment for a long holding period given that inflation ordinarily runs 2-3% annually, and can run hot at 4-5% for periods of time.  Add in taxes, and buying a long term bond today for a long term holding period is a loosing bet.  On the other hand, stocks have an earnings yield of ~4.2% based on the current price and ttm earnings.   Furthermore stocks can grow.  So the numerator in the earning yield calculation is actually getting larger.  Earnings growth is around 20% right now.  Will it always stay this high? No, it can go down.  But stocks are better priced now to deliver returns for a long term holding period.  And when looking at historic data, they have in the past performed much better than bonds over long holding periods.  Now, if for some reason bonds were yielding 11% and stocks had a PE ratio of 30 (earnings yield of 3.33%), I would be able to change my mind.  But thats not the environment we are in.  The reality is that stocks have historically been a better long term investment than bonds, and they are currently priced to deliver better performance than bonds going forward over a long term holding period.

Is it meant to imply that during the distribution phase it might make sense to change allocations as the relative strength of assets changes?

I guess I'm struggling a bit with the juxtaposition of using stock's historic data to dismiss conclusions drawn from comparing it to other asset classes' historic data (and those of MPT) and the juxtaposition of saying that only current figures are relevant with the justification that historically stock has done the best. Maybe I'm just not seeing the questionable assumptions in MPT.

And like the above religion comment implies, it does very much seem to come down to what bits you choose to accept and what you choose to question. I just want to make sure I know what implicit assumptions I'm choosing to accept or reject before I start travelling down a path because, like religion, it won't work if times get hard and you don't believe.

[SeattleCPA]

those graphs would be better if it showed the range of outcomes - i know its worst and best - but i think more outcomes skew towards the center and above than towards the lower or worst outcome situation.

Good point. I agree. (It's a bit of work to march cFIREsim through the calculations, export the numbers to Excel, arrange the data so you can plot it in a chart, plot, etc.... but what you suggest would definitely be more useful.)

[sisto]

I totally agree with @L.A.S. and @boarder42 on this. Almost all of your money is almost always going to be held for a long period of time. I hold 98% in stocks and plan to do so into retirement. I plan to protect myself against sequence of returns risk by holding some cash and adjusting my withdrawal rate if necessary. I can take less in down times and use some cash followed by bolstering the cash reserves when the market is crazy good.

[boarder42]

number 5 on value investing can be seen readily with what the market has done since 2008 - the reason stocks and bonds were "a good match" was because of typical negative correlation - but over the past 10 years they have actually both gone up in value and dropped in value almost in sync with each other.

[SeattleCPA]

However, those who don't follow MPT but take a more value oriented approach will look to the valuation asset as they are currently buying it as well as using history as a guide.  For instance, right now long term treasury bonds are paying about 3% $30 annually for each $1000 invested.  And as it is fixed in come, the $30 will not grow.   I consider this an insanely risky investment for a long holding period given that inflation ordinarily runs 2-3% annually, and can run hot at 4-5% for periods of time.  Add in taxes, and buying a long term bond today for a long term holding period is a loosing bet.  On the other hand, stocks have an earnings yield of ~4.2% based on the current price and ttm earnings.   Furthermore stocks can grow.  So the numerator in the earning yield calculation is actually getting larger.  Earnings growth is around 20% right now.  Will it always stay this high? No, it can go down.  But stocks are better priced now to deliver returns for a long term holding period.  And when looking at historic data, they have in the past performed much better than bonds over long holding periods.  Now, if for some reason bonds were yielding 11% and stocks had a PE ratio of 30 (earnings yield of 3.33%), I would be able to change my mind.  But thats not the environment we are in.  The reality is that stocks have historically been a better long term investment than bonds, and they are currently priced to deliver better performance than bonds going forward over a long term holding period.

Is it meant to imply that during the distribution phase it might make sense to change allocations as the relative strength of assets changes?

I guess I'm struggling a bit with the juxtaposition of using stock's historic data to dismiss conclusions drawn from comparing it to other asset classes' historic data (and those of MPT) and the juxtaposition of saying that only current figures are relevant with the justification that historically stock has done the best. Maybe I'm just not seeing the questionable assumptions in MPT.

And like the above religion comment implies, it does very much seem to come down to what bits you choose to accept and what you choose to question. I just want to make sure I know what implicit assumptions I'm choosing to accept or reject before I start travelling down a path because, like religion, it won't work if times get hard and you don't believe.

Here ya go...

MPT assumptions:

1.) risk it synonymous with volatility; that they are the same thing and volatility can be substituted for risk in MPT's calculations
2.) The markets are always efficient and immediately incorporate into the market price all known information either in the strong (perfect) or weak (essentially) sense, causing a financial asset to have one true value at any point in time which is it's market price.
3.) that the future volatility for an asset will be essentially the same as past volatility for that asset
4.) that past returns of an asset class are a reasonable predictor of the future returns +/- the volatility inherent in an asset
5.) that the future level of correlation between two assets will be the same as past correlation between those two assets
6.)  That the ideal portfolio is one which attempts the highest risk-adjusted (i.e. volatility adjusted) returns for an investor's desired risk-tolerance (i.e. volatility-tolerance).  An investor does this looking to past data, e.g. by taking into account past returns, past volatility, and past correlation between assets.

Value oriented investing premises-
1.) risk is the likelihood of loosing money over an expected holding period after taking into account any taxes, fees, and inflation.  Essentially risk is the likelihood of loss of real purchasing or consumption power over the holding period.
2.) That markets are often efficient in the long run....but not always in the short run.   At any given point in time a financial asset has two values, (1) the market price, and (2) the intrinsic value.  Intrinsic value is never known exactly.  But, value investing is attempting to determine an approximate intrinsic value for an asset (or asset class like an index fund) and then buying when the market price represents a good value with respect to intrinsic value.  This forms a safe investment since a value investor paid the market price but got the intrinsic value. It is risky to  pay too high of a market price for too little intrinsic value.
3.) That past contemporaneous rapid drops in prices (price drops due to volatility) make the present purchase of an asset safer.  After a drop in price there is a much better ratio of market price to intrinsic value since intrinsic value is usually much more stable than market price.  The investor is much less likely to overpay and the asset is more likely to revert towards the intrinsic value than not.
4.) Past returns can be used as a guide to see how an asset has behaved in the past under similar circumstances and market conditions, and how it might behave in the future, but an investor must also look to the price and try to ascertain intrinsic value of the assets they are buying in order to determine what the expected future return of an asset class is (e.g. yield on bonds, P/E ratios on stocks, book value, etc.).  An investor must be satisfied that the asset as priced promises reasonable safety over their holding period.
5.) Past correlation between asset is not fixed and can't really be counted on for anything.  Sometimes things move in sync and sometimes they are not synced, sometimes they move opposite.
6.) An investor reduces risk and constructs the ideal portfolio for themselves by seeking the best assets to attempt to maintain their purchasing power over their holding period, or holding periods, and training themselves to respond with equanimity to any price fluctuations and market gyrations that this entails.

/quote]

I wonder if your descriptions of what MPT is and how MPT works are really what people use in practice.  I don't think, for example, that your MPT points #3, #4, and #5 are really anything people believe, do they? Also, your point #5 for value investing is an important caveat Vanguard makes when they talk about the practical implementation of MPT in real life.

I wonder therefore if we're not really debating or discussing the same topic. It may be that we don't actually disagree. (E.g., I wouldn't agree with something that depends on returns, standard deviations and correlations being stable because they objectively aren't.)

FYI I provided my practical, working descriptions earlier. And what at a practical level MPT means to me is try to assemble portfolios of less the perfectly correlated asset classes. Be sure you understand the volatility in your portfolio if it matters to you... and be sure you understand that volatility will sometimes matter not just in terms of impact on investor behavior... but on other stuff too.

[Indexer]

MPT assumptions:

1.) risk it synonymous with volatility; that they are the same thing and volatility can be substituted for risk in MPT's calculations
2.) The markets are always efficient and immediately incorporate into the market price all known information either in the strong (perfect) or weak (essentially) sense, causing a financial asset to have one true value at any point in time which is it's market price.
3.) that the future volatility for an asset will be essentially the same as past volatility for that asset
4.) that past returns of an asset class are a reasonable predictor of the future returns +/- the volatility inherent in an asset
5.) that the future level of correlation between two assets will be the same as past correlation between those two assets
6.)  That the ideal portfolio is one which attempts the highest risk-adjusted (i.e. volatility adjusted) returns for an investor's desired risk-tolerance (i.e. volatility-tolerance).  An investor does this looking to past data, e.g. by taking into account past returns, past volatility, and past correlation between assets.

...

LAS, I think you see it as MPT VS value investing. They aren't mutually exclusive, and they don't contradict each other. MPT is a tool, not a religion. MPT is a tool to compare investments, asset classes, and portfolios based on their risk VS return trade off. It's just trying to create a measuring stick where there wasn't one. I also think you might be confusing Efficient market hypothesis with MPT. They are different things.

1) Risk according to MPT is the risk that you get returns different from what you expect. You expected 10%, and got 8%, that is a risk of lower than expected returns. Volatility is one way to measure how likely your returns will be different. Volatility tends to go both ways, up and down. If you're worried about getting returns less than what you expected, volatility isn't a bad thing to look at. Most formulas in MPT are comparing an investment, asset class, or an entire portfolio to a hypothetical risk free option. Is the extra return over the risk free rate worth the extra volatility that was added? Markowitz theorized that was a good way of taking something incredibly complicated(at the time) and converting it into a useful measuring stick.
2)  You are talking about Efficient Market Theory here, not MPT. None of this applies to MPT.
3)  It's a tool, not a religion. The person doing the calculation can substitute a different number. In most cases all you will have is past data, so yes, that does get used a lot.
4)  Again, it's a tool, not a religion. The person doing the calculation can substitute a different number.
5)  It's a tool. You can apply common sense. Example: US Stocks and US treasuries tend to move in different directions during crisis events thanks to flight to quality. If that was called into question you could use a different number.
6) Most investors don't like to see their portfolio drop 50%. Many investors would panic in that event and make impulsive(stupid) decisions. If a more moderate portfolio can achieve similar returns with substantially less downside then those investors will likely be better off with the moderate portfolio. Depending on when you need the money you might want to avoid big drops for logical reasons as well.


Value
1)  You need to stay above inflation? MPT wants you to stay above the risk free rate. Just plug inflation in as your risk free rate, which many investors do, and you are describing the same thing.
2)  Okay. None of this conflicts with MPT.
3)  Okay. I agree with everything you said. I like value investing and I like MPT as a tool. None of this conflicts with MPT. Again, I think your complaint is with EMH, not MPT.
4)  Okay. Buy things that are likely to give you higher returns without taking on substantial risk. How does this conflict with MPT?  MPT actually has calculations that help you measure how well you are doing this VS a benchmark of your choosing.
5) It's not a religion. Use common sense and readily available data. If facebook and Telsa both go down on the same day, it's likely chance or they are both moving with the overall market. If VOO and SPY both move in the same direction 100% of the time... it might be because they are tracking the same index. Is there a cause and effect relationship? No, then it's chance. Yes, how strong is that relationship and will it continue? Again, MPT is a tool to be used. It's not a religion. If there is a cause and effect relationship, use it as long as it makes sense.
6)  To summarize what you said: an investor builds a portfolio based on their income needs while limiting the amount of risk they are taking. You are constructing a portfolio with the understanding that you want to earn money but you don't want to take on excessive risk. That's the basis of MPT...

[CorpRaider]

To me, in addition to the other reasons stated, bonds are included because they save your bacon in a deflationary crash.  In real terms, treasuries whooped some ass during the great depression.  ALWAYS KEEP IT REAL.

Series I Bonds protect you from both inflation and deflation, just saying, as the rate floats with inflation but can't go below zero and the face amount is fixed (although it wouldn't be as good as notes in a 1929, '37, or 2008, scenario).  I have read credible arguments that one should not put a single $ into other fixed income instruments until they have maxed out their annual allocation of I Bonds. 

I know a dude who wrote a post about it one time. 

[TomTX]

To me, in addition to the other reasons stated, bonds are included because they save your bacon in a deflationary crash.  In real terms, treasuries whooped some ass during the great depression.  ALWAYS KEEP IT REAL.

Series I Bonds protect you from both inflation and deflation, just saying, as the rate floats with inflation but can't go below zero and the face amount is fixed (although it wouldn't be as good as note in a 1929, '37, or 2008, scenario).  I have read credible arguments that on should not put one $ into other fixed income instruments until they have maxed out their annual allocation of I Bonds. 

I know a dude who wrote a post about it one time.

Yep, I'm quite fond of the I-Bond. 

[SeattleCPA]

MPT is a tool, not a religion.

Well said, indexer. Well said.

[CorpRaider]

Harry Markowitz definitely appears to think it is somewhat less than a religion. 

First, dude said yeah, I came up with the efficient frontier, but when it came time to allocate real money I just did 1/n.  Then he's quoted (in the last month) making allocation/market calls based on the hurricanes and reconstruction of Houston and Puerto Rico.  haha!

[ChpBstrd]

I don't believe in MPT.  I view MPT is a mathematical religion constructed around the belief that risk = volatility.  Unfortunately, for most folks who are not short term traders, volatility != risk. 

Risk is defined as the likelihood that the purchasing power of an investment after all taxes, fees, and inflation will not be maintained over the holding period.

On a short term time scale used by traders: say a matter of weeks, months, or up to a year or two, volatility is a good approximation of risk, because swings in the asset price will create a very real risk of the asset failing to maintain its purchasing power over the short time frame.

However if someone has 10 or 20 or 30 years to invest, volatility is less meaningful.  What is far more risky is that inflation, taxes, and fees will consume too much of the real value of an asset, or of an entire portfolio leaving the investor with less purchasing power at the end of the 10, 20, or 30 year holding period.

Take for instance a current 10 year bond yielding under 3%.  I would say this is an incredibly risky investment to buy and hold for 10 years.  The risk of loosing real, net purchasing power is virtually 100%.  Whereas a diversified group of stocks will have a much higher likelihood of maintaining (and most likely increasing) purchasing power.  I think the differences get even more stark with 20 year bond vs. stocks, or 30 year bond vs. stocks.

So why do we have MPT?  I think we have it because a bunch of not-actually-rich academics were able to produce some beautiful equations and charts and graphs when they substituted (past) volatility for true risk and looked at historical data.  Greek letters could be invoked, people could have equations and ratios named after them.  Some could even win Nobel prizes and thereby become anointed.  It has become an academic's nirvana.  When things get wonky, they show that the math still holds if the event is considered exceptionally rare....  The joke goes, "It's more convenient to add a sigma, than to change a formula."

I would argue in fact that the entire financial crisis of '08/09 was exacerbated by the use of MPT to the exclusion of common sense. One idea comes to mind which caused people to ignore enormous risk, namely  VAR.

VAR stands for Value at Risk, and it must be computed by banks regularly to determine how much "risk" is in a portfolio. But it is a statistical method of computing risk based on passed volatility.  So for example a collection of shit-loans sub-prime loans were rated investment grade and have very low statistical risk, whereas a junk-bond that has already dropped like a rock and was rated below investment grade but was trading below a conservative estimate of available assets of the firm would be considered "risky."  The result is that the purchase of the statistically less risky assets were allowed to be leveraged whereas the purchase of the statistically more risky (but actually less real risk because it is backed by real assets) would likely have to be hedged against.  Some risk management system.

If asystemic risk can be better mitigated through securities analysis than through diversification, wouldn't that lead one to stock-picking?

If volatility is a poor measure of a stock's risk, how about leverage? (They are highly correlated anyway)

Also, isn't any factor that could be analyzed subject to the risk of changing? E.g. the junk bond for a highly leveraged company with valuable assets and a solid business might be a good deal until the next quarter when their new product flops, an executive scandal breaks out, a parter sues them, suppliers cut them off, earnings disappoint, they announce an unforeseen write-off, or an accident occurs. The same analysis then finds the bonds to be a bad deal, but you already bought them.

Re: academics and MPT - is this an ad hominem argument?