Portfolio Design: Idiots v. Gurus

[Radagast]

For a long time I have noticed that naively filling every asset class (or every second, third, or fourth asset class) often gave better results than most "expert" portfolios using both https://portfoliocharts.com/ and https://www.portfoliovisualizer.com/. Here is my formal investigation into it.

Rules: I equally weighted every or every 2nd, 3rd, or 4th asset class using both withdrawal types (safe and perpetual) at https://portfoliocharts.com/portfolio/portfolio-matrix/. Spare change went to cash, or was taken away from cash, to keep all other assets equally weighted. This was a form of sandbagging my hypothesis, because Cash is Trash as Portfolio Charts confirms.

Note 1: I like to use withdrawal rates best of all backtesting objectives. Are long term returns more important, or is stability more important? With safe withdrawal rate there is no bullshitting about one or the other is more important: if it was more important why didn't it result in a higher safe withdrawal rate?

Note 2: There are really only let's generously say 3 data points affecting the results here. All of the portfolios except Japan had their worst case in 1970-1974. Some of the countries give slightly different results so lets call that a second effective datapoint, as we can throw cold water on some of the weakest contenders. Pay special attention to Japan, as that is the only true out of sample datapoint for most "expert" portfolios.

Note 3: Portfolio Charts has 9:10:4:3 HomeStocks:ForeignStocks:Bonds:"Real"Assets, so all naive portfolios follow that basic ratio, which might be significant. I did on odd occasions throw out TSM and International TSM for simplicity.

Note 4: SWR is determined based on the single worst year in the record for 30-year rolling returns. Average returns would make a very different ranking.

Here are the results. Naive portfolios are white, "experts" get colors. The withdrawal rate is listed for each, could be some typos but the rankings should be correct. I am surprised to note that even with cash sandbagging, the "experts" mostly demonstrated no skill and underperformed naive asset class selection. Only a few expert portfolios appear to show skill, including the Ivy, 7Twelve, and Pinwheel. Even the 7Twelve and Pinwheel portfolios seem to be more about extreme diversification than skill, with 12 and 8 slices. Perhaps a future data point will show that the Ivy Portfolio was just lucky? The average "expert" portfolio looks to have underperformed the average naive portfolio.


And here is what went into the naive portfolios. TMI I know.


If I ever make my own portfolio I will call it the "I'm an Idiot Portfolio" :D

Thanks mjr! Had to run just as I posted, not sure what went wrong.

[Tyler]

A post like this deserves way more attention.  Nice work!

Note 2: There are really only let's generously say 3 data points affecting the results here. All of the portfolios except Japan had their worst case in 1970-1974. Some of the countries give slightly different results so lets call that a second effective datapoint, as we can throw cold water on some of the weakest contenders. Pay special attention to Japan, as that is the only true out of sample datapoint for most "expert" portfolios.

Note 3: Portfolio Charts has 9:10:4:3 HomeStocks:ForeignStocks:Bonds:"Real"Assets, so all naive portfolios follow that basic ratio, which might be significant. I did on odd occasions throw out TSM and International TSM for simplicity.

I think these are very important points.  You may call your selection methodology "naive", but I'd argue that the built-in diversification with an eye on Japan is a very intelligent approach.  My big takeaway is not that the various gurus are unskilled, but that perhaps they should pay more attention to home country bias and the benefits of true asset diversification rather than simple stock tilts.

[kenmoremmm]

where do i send the check for all of this work?

after compiling these data, did it change your investment strategy?

[BicycleB]

Finally PTF. For the first time ever.

Quite intrigued.

[Radagast]

A post like this deserves way more attention.  Nice work!

Note 2: There are really only let's generously say 3 data points affecting the results here. All of the portfolios except Japan had their worst case in 1970-1974. Some of the countries give slightly different results so lets call that a second effective datapoint, as we can throw cold water on some of the weakest contenders. Pay special attention to Japan, as that is the only true out of sample datapoint for most "expert" portfolios.

Note 3: Portfolio Charts has 9:10:4:3 HomeStocks:ForeignStocks:Bonds:"Real"Assets, so all naive portfolios follow that basic ratio, which might be significant. I did on odd occasions throw out TSM and International TSM for simplicity.

I think these are very important points.  You may call your selection methodology "naive", but I'd argue that the built-in diversification with an eye on Japan is a very intelligent approach.  My big takeaway is not that the various gurus are unskilled, but that perhaps they should pay more attention to home country bias and the benefits of true asset diversification rather than simple stock tilts.

Thank you for bumping! I was hoping for discussion but it's bad form to bump my own post, I was getting scared the whole thing would be a one post wonder.

I agree that 1/n weighting in general is more sophisticated than naive. Both for individual stocks and for portfolio components I have read research shoing that 1/n is surprisingly effective, for example people in retirement plans who simply apply equal weight to every available option. Provided the expense ratios are low, that is actually a very good strategy.

You are also probably thinking that including a "real" asset is also something many "gurus" miss, and I am coming around to that line of thought, especially as it did not seem to hurt the Japan case, provided they are used in moderation. In fact I think the Portfolio Charts weighting which is about 40% home country stock*, 36% foreign stock, 16% bonds cash, 8% commodities is probably a pretty ideal asset allocation. *Depending how you count REITs. Take the Merriman Ultimate Buy and Hold: if it had diverted a percentage of its bonds to commodities it would have been one of the best portfolios out there (at the cost of even more slices).

One thing I notice is that guru portfolios do best in Japan and Germany, which I expect is a result of heavy bond allocations. Relative to the English speaking countries, I believe bond yields on those two started higher, and are now much lower. This created a huge tailwind for gurus who favored huge helpings of  long term and intermediate bonds, but did no favors to Bernstein who strongly prefers large amounts of short term bonds. I expect this to reverse at some point, and for that reason I think gurus who have outsized allocations to either extreme are poorly diversified.

But I am not convinced now that any guru is better than randomly selecting a variety of funds with relatively low expense ratios.

[Radagast]

where do i send the check for all of this work?

after compiling these data, did it change your investment strategy?

No check, like most posts this was for self confirmation, vindication, and gratification! Often people post a list of funds and ask what to invest in, and forumers would say things like "76.2% S&P500, 14.2% midcap, 9.6% small cap will give you the equivalent of total market which is what you want" while I would be like "cut the precision, you have large, mid, small, bond, and international funds with low expense ratios, just equally weight those and call it a day, it's not like anyone has any idea anyhow." So I am basically just trying to prove myself right.

No changes, I put these together in a few hours. My strategy don't turn on no dime. It did influence me in two ways though. 1st, I have always been compelled by the zen of just putting it all in total world stock market and saying "ooohhhmmmm." However I notice that allocations with fewer than four slices often make trips to the bottom, and in some cases live there permanently. All of the naive allocations had at least 7 slices plus cash and did very well. I have increasingly been thinking that 7-15 slices might be better although many people would call that crazy, especially if spread around nine accounts. 2nd, while there is only one solid data point for gold/commodities, it did not really have a significant downside anywhere (in moderation) and the Ivy Portfolio did great for Japan in the out-of-sample point. I am warming toward a 8%-16%ish allocation to those.

[Radagast]

Also big thanks to mjr. For some reason my original links did not embed the images, and I had to leave for a few hours just as I posted. mjr made working image links and then deleted the post.

[effigy98]

Interesting. Max diversification looks like the most consistent winner.

[sailinlight]

Is there a dummy version of the results of this? It seems that the takeaway is that for a 30 year retirement, Golden Butterfly affords the largest withdrawal rate (or the more useful converse 'greatest chance of not running out of money').

[vand]

Diversification is the only free lunch in investing.

All this shows that there are many successful ways to build a portfolio. It may not be an overexaggeration to suggest that the "naive" approach is to think that a larg concentration in stocks is the One True Way to wealth accumulation.

[effigy98]

What exactly happened in Japan to the Golden Butterfly? Is there not enough deflation protection?

[Kierun]

whut?  can someone dumb this down for me as to what it I'm looking at here? looks interesting but I'm an idiot and don't understand, all I get is...colors.

[BicycleB]

@Kierun, I think it's working like this:

Bottom section (the part with all the little black blocks):
Each sector under a yellow label, such "0th" (zero-th) or "5th" (fifth) shows the allocation of investments in one of the "idiot" or "naive" portfolios that the OP made. Each little block shows an investment type, and the number inside the block is the % of the portfolio that went into that investment type. So the first or "0th" example, in the Domestic or USA row, the left-most investment block is TSM (which stands for Total Stock Market); TSM had 4% of the portfolio invested in it. You can go through all the rows in the portfolio, and each portfolio's numbers should add up to 100%. The abbreviations (TSM for Total Stock Market, LCB for Large Cap Blend or something like that, etc) are explained in the website that the OP used to make all these calculations (portfoliocharts.com), which has a ton of calculators based on historical investment data.

Top section:
The investment results of the idiot portfolios are shown, along with the investment results of some other, more carefully crafted portfolios recommended by investment "experts" such as the authors of various investment books. The color coded boxes show the results of the experts' portfolios, referred to as gurus in the thread title; the white boxes show the results of the "idiot" aka "naive" portfolios, where the OP basically picked investments using various random methods (his choices were detailed in the bottom section of the colorful post). Each colored box in the top section gives the common name of the expert portfolio that it describes. Each white box refers to bottom section - "1st naive" in the white box means the portfolio labeled "1st" in the bottom section; the portfolio header "1st" is yellow in the bottom section.

The plain red boxes in the top section refer to portfolios that were one of the worst two portfolios in some country (any country). So the red portfolios were historically risky for some investors; implicitly, perhaps a red flag. The other colors show expert portfolios that escaped the scary cellar of these league standings, so to speak. The expert portfolios are each described in detail on portfoliocharts.com, which Tyler the Mustachian created as his gift to the world for free portfolio calculations based on historical data.

Each box in the top section ends in a number giving the highest withdrawal rate that worked in all investment years for that portfolio during the period where data could be found (I think 1970 to present) - in other words, if you started by investing in the worst possible year, the box shows the investment rate you could have used. I think that this required 30 year periods though, so years that started less than 30 years ago may not be included. Perhaps investors who started in 2000, for example, who suffered from the tech bust of 2001-2003 and then the 2008-09 Great Financial Crisis, will have lower withdrawal rates than shown.

I think the "Safe" columns meant "didn't run out of money for 30 years at that rate". You can see that for each country, the portfolios are in order from highest withdrawal rate to lowest, starting at the top. The "Perpetual" columns show the rate where the portfolio ended the test period with as much money as it started; the perpetual rate is how much you could withdraw per year historically and still end up with the original investment value. You can see that's a lower rate than the so-called safe rate, but it would let your investments last longer, so it's good to know for people who have longer investment timeframes, such as people retiring at 35 and living to 90. For them, the perpetual rate might actually be safer!

Clear as mud, eh??

[Radagast]

Is there a dummy version of the results of this? It seems that the takeaway is that for a 30 year retirement, Golden Butterfly affords the largest withdrawal rate (or the more useful converse 'greatest chance of not running out of money').

Not really :-| in general I am saying that explanations of why past stock, bond, and "real assets" did what they did might not be correct so much as they are good stories. As Vand says, diversification is the only free lunch.

My asset allocation advice (which I am tweaking slightly as a result of this) is pretty simple:
At least 50% of total assets in stocks
At least 25% but not more than 50% of total allocation in your home country/currency stocks (or other business)
Meaningful allocation to international stocks (unhedged)
Include 15-40% bonds
Include a "real asset," and yes your home or rental real estate count.
In fact investment-grade real estate would be fine for 80% of this
Plus lots of caveats!

Like I said in OP, Japan was the only true test of the Golden Butterfly. Performance in the US was dominated by a single year, 1973, which doesn't give much information going forward. My explanation of why a single data point is not very useful.here. Also gold behaved oddly in 1970-75 because over that period the government started off with price controls (at a very low price btw), and then gradually released it, so it isn't a particularly useful data point. You would have to experiment with substituting it with REITs or commodities or even cash to get a non-interventionist picture.

What exactly happened in Japan to the Golden Butterfly? Is there not enough deflation protection?

Both the largest stocks in Japan and the smaller value companies did poorly in that period. Gold was in the dead center of a two decade bear market when the Japanese needed it most in 1990. Cash did its usual thing and went nowhere. That left the 20% allocation to long term bonds to carry the entire thing. You might say it did not have enough deflation protection, but I say it did not have enough diversification ;). A 6th equally weighted slice to Total International Stock Market would have brought the Japan case way up, and if the 6th slice was International Small Cap Value instead it would have backtest as the best portfolio in the English speaking world in addition to being near the top in Japan and Germany.

[Radagast]

I think it's working like this:

Bottom section (the part with all the little black blocks): Each sector under a yellow label, such "0th" (zero-th) or "5th" (fifth) shows the allocation of investments in one of the "idiot" or "naive" portfolios that the OP made. Each little block shows an investment type, and the number inside the block is the % of the portfolio that went into that investment type. So the first or "0th" example, in the Domestic or USA row, the left-most investment block is TSM (which stands for Total Stock Market); TSM had 4% of the portfolio invested in it. You can go through all the rows in the portfolio, and each portfolio's numbers should add up to 100%. The abbreviations (TSM for Total Stock Market, LCB for Large Cap Bonds or something like that, etc) are explained in the website that the OP used to make all these calculations (portfoliocharts.com), which has a ton of calculators based on historical investment data.

Top section: The investment results of the idiot portfolios are shown, along with the investment results of some other, more carefully crafted portfolios recommended by investment "experts" such as the authors of various investment books. The color coded boxes show the results of the experts' portfolios, referred to as gurus in the thread title; the white boxes show the results of the "idiot" aka "naive" portfolios, where the OP basically picked investments using various random methods (his choices were detailed in the bottom section of the colorful post). Each colored box in the top section gives the common name of the expert portfolio that it describes. Each white box refers to bottom section - "1st naive" in the white box means the portfolio labeled "1st" in the bottom section; the portfolio header "1st" is yellow in the bottom section. I think that all the plain red boxes refer to portfolios viewed as generic, such as all stock or 60% stock 40% bonds, while other colors refer to specific expert portfolios. The expert portfolios are each described in detail on portfoliocharts.com, which Tyler the Mustachian created as his gift to the world for free portfolio calculations based on historical data.

Each box in the top section ends in a number giving the highest withdrawal rate that worked in all investment years for that portfolio - in other words, if you started by investing in the worst possible year, the box shows the investment rate you could have used. I think the "Safe" columns meant "didn't run out of money for 30 years at that rate". You can see that for each country, the portfolios are in order from highest withdrawal rate to lowest, starting at the top. The "Perpetual" columns show the rate where the portfolio ended the test period with as much money as it started; the perpetual rate is how much you could withdraw per year historically and still end up with the original investment value. You can see that's a lower rate than the safe rate, but it would let your investments last longer, so it's good to know for people who have longer investment timeframes, such as people retiring at 35 and living to 90.

Thanks BicycleB that is completely correct! Well except one minor change: I used red for any portfolio which made it to the bottom 2 rows for any country, which is why the "Larry Portfolio" is red, and also the Bill Bernstein "No-Brainer."

[Radagast]

As a side note there might be other ways to accomplish a higher safe withdrawal rate beside a static asset allocation which are commonly discussed around here. For example, the reverse glide path using a 5-year CD ladder, or a variable withdrawal rate.

[BicycleB]

@Radagast, thanks for answering.

I have a lot of boxes that appear on my computer as red. In the USA Safe column, is the red for Larry Portfolio supposed to be the same as the color for the bottom three fields in that column?

ETA: Never mind, I now realize what you said. Read means bottom of ANY column - as in, historically risky somewhere! Very thoughtful. Thx, I can edit my explanation post now.

ETA: Ok, main explanation post edited.

[Radagast]

@Radagast, thanks for answering.

I have a lot of boxes that appear on my computer as red. In the USA Safe column, is the red for Larry Portfolio supposed to be the same as the color for the bottom three fields in that column?

ETA: Never mind, I now realize what you said. Read means bottom of ANY column - as in, historically risky somewhere! Very thoughtful. Thx, I can edit my explanation post now.

Yep, to my mind picking future winners is hard, but avoiding the worst outcomes is probably possible. To bad about the "No Brainer" though, if it had used intermediate term bonds and small value stocks instead of just small it would have been notably higher. Even the "Larry Portfolio" would probably have been better in real life in the UK, as it dedicated half its 70% bond allocation to inflation protected bonds.

[Tyler]

Each box in the top section ends in a number giving the highest withdrawal rate that worked in all investment years for that portfolio during the period where data could be found (I think 1970 to present) - in other words, if you started by investing in the worst possible year, the box shows the investment rate you could have used. I think that this required 30 year periods though, so years that started less than 30 years ago may not be included. Perhaps investors who started in 2000, for example, who suffered from the tech bust of 2001-2003 and then the 2008-09 Great Financial Crisis, will have lower withdrawal rates than shown.

Normally you're correct about 30-year SWRs for periods that started less than 30 years ago, but I actually figured out a way to account for that without trying to predict the future. You can read about it here: https://portfoliocharts.com/2017/03/21/how-to-predict-withdrawal-rates-without-a-crystal-ball/

[vand]

As Vand says, diversification is the only free lunch.

Haha, you probably know that they're not my words... I'm just requoting Harry Markowitz, whose work I believe puts him firmly in any list of the top 10 most influential investors of the 20th century alongside the likes of Buffett and Bogle.  If today's major trend is low cost passive indexing, I think the next big trend in financial markets may well be the rise of multi-asset diversification.  This may well come about when the current paradigm eventually morphs into something more hostile for traditional portfolios.

[effigy98]

What exactly happened in Japan to the Golden Butterfly? Is there not enough deflation protection?

Both the largest stocks in Japan and the smaller value companies did poorly in that period. Gold was in the dead center of a two decade bear market when the Japanese needed it most in 1990. Cash did its usual thing and went nowhere. That left the 20% allocation to long term bonds to carry the entire thing. You might say it did not have enough deflation protection, but I say it did not have enough diversification ;). A 6th equally weighted slice to Total International Stock Market would have brought the Japan case way up, and if the 6th slice was International Small Cap Value instead it would have backtest as the best portfolio in the English speaking world in addition to being near the top in Japan and Germany.

Ahhh yes, that would have helped a lot. Based on the Dalios drum beat for emerging, probably the best place vs too much developed to have less correlation with US market.

[Kierun]

@BicycleB Many mahalos!  It all makes so much more sense now.  And thanks @Radagast for all that info. 

[ChpBstrd]

I wish someone could analyze Nassim Taleb’s black swan portfolio, which consists of a huge allocation to treasuries plus a small allocation to long call options on the stock market. I wouldn’t expect it to do amazingly well, but it would definitely avoid major damage during corrections and apply upside-only leverage in recoveries. It’s too bad we have no easy way to measure it because historical options pricing is so hard to find or calculate.

What I really wish someone could analyze is a protected put or collared portfolio, which is another way to avoid big drawdowns. I suppose you’d have to build a spreadsheet options pricing model and feed it with historical volatility data, interest rates, etc. The slightest inaccuracy, such as misestimating theta or gamma, would throw it completely off.

[ardrum]

I don't have it open right now, but doesn't portfoliocharts show the highest SWRs with something like a 2:1 ratio of US small cap value to international small cap/small cap value with a smidge (5-10%) of gold?

[BicycleB]

^Even if the data show that for the period described, how likely is it that such a precise portfolio will continue its outperformance?

AFAIK, there's no logic-based reason why portfoliocharts.com limits its data to 1970 or so to the present. I think it's just that that's the era Tyler was able to find data for. Is that long enough of a period to distinguish between strategies that have enduring value, vs strategies that had a hot streak during that time?

Questions like these are why I find Radagast's results so intriguing. Maybe they suggest that diversification as a general principle, rather than any particular allocation, is what has enduring value.

Still pondering how to sort these things out.

[Tyler]

AFAIK, there's no logic-based reason why portfoliocharts.com limits its data to 1970 or so to the present. I think it's just that that's the era Tyler was able to find data for.

Correct.  One could argue gold data under the Bretton Woods system is irrelevant today or that very old stock data is sketchy due to company count in the sample, but neither drove the 1970 date on Portfolio Charts.  It's just the limits of what I can find for so many assets, and I'll happily expand that if/when I find reliable new sources. 

I will note that Eugene Fama and Ken French have done tons of research on the small and value premiums going back to 1927 with plenty of data and reasoning to back it up.  Larry Swedroe has also written prolifically on the topic.  I encourage anyone who would like to know more to think beyond the PC data and read about it for yourself before either dismissing it entirely or betting your retirement on it.  My numbers are just one voice, not the final word. 

I also generally discourage people from getting too crazy with seeking out the highest SWR for a concentrated portfolio of any one asset.  Be smart about it. All numbers aside, diversification is also your friend when it comes to risk management. 

[ardrum]

Great discussion and analysis.  It's hard to disagree with an approach grounded in broad diversification at minimal tax/cost.

I always find it fun to look at past data even though it's definitely no crystal ball.  I'd bet sticking to whatever allocation one goes with is a good way to stay out of the bottom of the list vs frequently jumping into and out of different allocations chasing results.

[Tyler]

Diversification is the only free lunch in investing.

Yep. Just as interesting as the variety of portfolios at the top of Radagast’s list is the remarkable consistency at the bottom.

[kenmoremmm]

is there a way to cleanly rank the portfolios from an overall perspective? visually, golden butterfly, ivy, or 7twelve or one of the naives looks to be best. can you capture the average SWR's  as measured across all countries on the list?

[chevy1956]

Diversification is the only free lunch in investing.

Yep. Just as interesting as the variety of portfolios at the top of Radagast’s list is the remarkable consistency at the bottom.

Here is the kicker. Will those portfolios on the bottom outperform over the next 20-50 years. I have real doubts being heavily invested in gold and bonds will perform well over that time period and those portfolios at the top appear to be heavy on gold and bonds. I have some bonds for safety but I'm not purchasing any gold. I also wonder if the world economy will continue to perform so poorly. I'm more than comfortable with having 50% of my portfolio in a world index tracker. In fact I think that this is hugely important based on being as diversified as possible within the equity market.

Diversification works but it's not just across markets but within markets.

[Radagast]

As a follow up, anyone care to guess how the naive portfolios did in terms of average returns? 'Cause I had some guesses and they totally fell short.

More notes:
Along with withdrawal rates, I also like to look at average returns. For one thing, the safe withdrawal rate is a single data point, meaning we can't draw very strong conclusions from it. Average still has the problem of being backward looking over a specific time period, but at least every year contributed equally to it. Also, it is good to get a perspective on how you will do in the 90% of the time when things are not going terribly.

These are arithmetic average returns, if I am reading this correctly. That would make them a little on the optimistic side for volatile portfolios because an 80% decrease has a far greater impact on your money than an 80% increase, which would not be reflected in the arithmetic average.

I never said it, but for the US "world stock" is 50% US/50% ex-US, which may give a rebalancing bonus. The other countries have a straight up global allocation.

The Pinwheel Portfolio and Ivy continue to shine, while Swensen, Merriman, and coffeehouse continue along the path of mediocrity which is probably good in terms of long term survival.

The Larry Portfolio has the distinction of being the only portfolio to be colored red for both safe withdrawal rate and average return. Whatever else you do, don't invest like this. 95% of the time this portfolio will have no upside, and 5% it has significant downside. I have always been mystified why this one exists because even a casual review of history shows government bonds have very little return and very little risk, until the issuing country gets into trouble and jacks up inflation while capping interest rates and the previously "safe" bonds switch to pure risk. The US did it in the 1940's, and other countries in this list did so to a much greater extreme.

So about those naive portfolios: don't forget they all have cash allocations. Cash is not a huge drawback for safe withdrawal rates, but it is a big bag of bowling balls for average returns.

Aaaannd I think it is pretty obvious that experts have no talent. They were already looking "meh" with the whole worst case withdrawal rates thing, but on average they are terrible.

Man I wish I could be this big an idiot. Is it still being naive if you try to act as if you were naive, or does the act of trying to act naive no longer make you truly naive?

Edit: fixed typo in image

[Radagast]

What exactly happened in Japan to the Golden Butterfly? Is there not enough deflation protection?

Both the largest stocks in Japan and the smaller value companies did poorly in that period. Gold was in the dead center of a two decade bear market when the Japanese needed it most in 1990. Cash did its usual thing and went nowhere. That left the 20% allocation to long term bonds to carry the entire thing. You might say it did not have enough deflation protection, but I say it did not have enough diversification ;). A 6th equally weighted slice to Total International Stock Market would have brought the Japan case way up, and if the 6th slice was International Small Cap Value instead it would have backtest as the best portfolio in the English speaking world in addition to being near the top in Japan and Germany.

Ahhh yes, that would have helped a lot. Based on the Dalios drum beat for emerging, probably the best place vs too much developed to have less correlation with US market.

I recall that international small cap value generally backtests better than emerging, but I agree that going forward there is no reason to expect that. When in doubt, do both?

[Radagast]

I wish someone could analyze Nassim Taleb’s black swan portfolio, which consists of a huge allocation to treasuries plus a small allocation to long call options on the stock market. I wouldn’t expect it to do amazingly well, but it would definitely avoid major damage during corrections and apply upside-only leverage in recoveries. It’s too bad we have no easy way to measure it because historical options pricing is so hard to find or calculate.

What I really wish someone could analyze is a protected put or collared portfolio, which is another way to avoid big drawdowns. I suppose you’d have to build a spreadsheet options pricing model and feed it with historical volatility data, interest rates, etc. The slightest inaccuracy, such as misestimating theta or gamma, would throw it completely off.

I am skeptical of Taleb's barbell strategies. How do you know which asset will turn out to be riskless and which will turn out to have the highest return? If you are wrong you could end up with either all downside or no upside. The Larry portfolio attempts this (without options) and look how it ended up. How does he know that walking and weight lifting are all you need, and running and biking are pointless? I prefer the smorgasbord approach.

That said I wish I knew more about options, but rounding to the nearest 1/10th of a percent, my knowledge rounds to 0.

[Radagast]

I forgot my most important thank you to @Tyler ! I have done a lot of backtesting but am a lightweight. Tyler put together the entire Portfolio Charts website with a gajillion backtesting methods and personally found a huge amount of data which would otherwise not be available at all. Portfolio Charts fills a huge gap between Portfolio Visualizer which has a million assets and includes actual funds but scarcely goes back to the year 2000, and CfireSim which goes to 1872 but with a narrow range of assets and only for the US.

[chevy1956]

It's interesting but it looks like some of the portfolios with a higher SWR have lower returns and vice versa. This is why I'm skeptical of the portfolios that are heavy in gold and cash type assets (bonds, treasuries, etc). Sure we might end up with poor returns in equities markets that result in those portfolios doing well that have gold and cash type assets but there is no guarantee that this will occur.

I like simple and I suppose naive portfolios foot the bill there.

It will be interesting how this changes over the next 50 odd years.

[SotI]

Just posting to say "thanks"! This is quite some food for thought.
I find it also interesting how relatively poorly the "simple" Total Stock or Classic 60:40 have been performing.

My portfolio is regionally diversified, but so far I have not included my home property or REITs in it.

Diversification is the only free lunch in investing.

Yep. Just as interesting as the variety of portfolios at the top of Radagast’s list is the remarkable consistency at the bottom.

Here is the kicker. Will those portfolios on the bottom outperform over the next 20-50 years. I have real doubts being heavily invested in gold and bonds will perform well over that time period and those portfolios at the top appear to be heavy on gold and bonds. ...

I am wondering, too, considering FED/ECB policy over the last decade (from a European pov, I am particularly looking at Germany, UK and even Japan here for comparison).

[vand]

You can learn so much about investing from the myriad of good blogs (and books) available. Tyler's Portfoliocharts is just one example, but there are lots out there.

While I fully respect and enjoy MMM but it's not really an Investing blog, per se. Your universe becomes overly narrow if all you read is FIRE blogs which multiply the same lines of thinking when it comes to personal finances and how to invest.

For example I'm currently working my way through https://ofdollarsanddata.com/

Heer is an excellent post he wrote on the subject of AA and portfolio efficiency:

https://ofdollarsanddata.com/where-to-invest-when-youre-investing/

Again, none of this should come as a surprise if you have already really studied in detail the concepts of portfolio theory, but few FI blogs tackle the topic of investing with this level of throughness (and of course, it could be argued, why should they).

None of this contradicts what the FIRE community already preach (he is highly scathing of the Active/Hedge Fund industry, for example), but its about expanding your knowledge, digging beneath simplistic headline numbers - ie acknowledging that "distribution of outcomes matters greatly" - and tailoring your investing accordingly.

[ChpBstrd]

It's interesting but it looks like some of the portfolios with a higher SWR have lower returns and vice versa. This is why I'm skeptical of the portfolios that are heavy in gold and cash type assets (bonds, treasuries, etc). Sure we might end up with poor returns in equities markets that result in those portfolios doing well that have gold and cash type assets but there is no guarantee that this will occur.

I like simple and I suppose naive portfolios foot the bill there.

It will be interesting how this changes over the next 50 odd years.

True. So much of this is history-specific. Since 1970, the US went off the gold standard, had a tech boom in equities, inflated real estate prices with government subsidized financing, and from the early 1980s to the present has had a bond rally of massive proportions. Meanwhile investors endured stagflation in the 70s, a tech bubble that burst in 2000, a real estate bubble that burst in 2008, and the worst recession since the 1930s. Will any of that happen in the next 50 years?

Selecting the optimal AA now would require knowledge of the future, not the past. Will there be a biotech boom driven by gene editing? Will bonds revert to their norms or go negative like in Europe? Will Fannie and Freddie Mac be broken up and sold off? Will the tech-driven attention economy hit a natural growth limit or see a backlash as their products become recognized as addictive? Will average savings rates continue to decline or are millennials naturally more frugal? Will there be a major war? Will blockchain disrupt banking and tank the dollar? How will demographic graying affect aggregate demand? Will the US remain a democracy? Will self-driving cars that we don’t own become the new transportation economy? Will parts of New Orleans, Miami, Charleston, Tampa, NYC, and Houston be abandoned and who pays for that? Will China continue growing, stagnate at some point, or collapse? What will be the average inflation rate for the next 3 decades?

If one knows any of this with certainty, portfolio selection becomes easy.

[chevy1956]

It's interesting but it looks like some of the portfolios with a higher SWR have lower returns and vice versa. This is why I'm skeptical of the portfolios that are heavy in gold and cash type assets (bonds, treasuries, etc). Sure we might end up with poor returns in equities markets that result in those portfolios doing well that have gold and cash type assets but there is no guarantee that this will occur.

I like simple and I suppose naive portfolios foot the bill there.

It will be interesting how this changes over the next 50 odd years.

True. So much of this is history-specific. Since 1970, the US went off the gold standard, had a tech boom in equities, inflated real estate prices with government subsidized financing, and from the early 1980s to the present has had a bond rally of massive proportions. Meanwhile investors endured stagflation in the 70s, a tech bubble that burst in 2000, a real estate bubble that burst in 2008, and the worst recession since the 1930s. Will any of that happen in the next 50 years?

Selecting the optimal AA now would require knowledge of the future, not the past. Will there be a biotech boom driven by gene editing? Will bonds revert to their norms or go negative like in Europe? Will Fannie and Freddie Mac be broken up and sold off? Will the tech-driven attention economy hit a natural growth limit or see a backlash as their products become recognized as addictive? Will average savings rates continue to decline or are millennials naturally more frugal? Will there be a major war? Will blockchain disrupt banking and tank the dollar? How will demographic graying affect aggregate demand? Will the US remain a democracy? Will self-driving cars that we don’t own become the new transportation economy? Will parts of New Orleans, Miami, Charleston, Tampa, NYC, and Houston be abandoned and who pays for that? Will China continue growing, stagnate at some point, or collapse? What will be the average inflation rate for the next 3 decades?

If one knows any of this with certainty, portfolio selection becomes easy.

This is exactly how I feel about portfolio theory. It's a guide. You can't predict the portfolios that will be successful in the future because you don't know what will happen in the future.

[dragoncar]

Each box in the top section ends in a number giving the highest withdrawal rate that worked in all investment years for that portfolio during the period where data could be found (I think 1970 to present) - in other words, if you started by investing in the worst possible year, the box shows the investment rate you could have used. I think that this required 30 year periods though, so years that started less than 30 years ago may not be included. Perhaps investors who started in 2000, for example, who suffered from the tech bust of 2001-2003 and then the 2008-09 Great Financial Crisis, will have lower withdrawal rates than shown.

Normally you're correct about 30-year SWRs for periods that started less than 30 years ago, but I actually figured out a way to account for that without trying to predict the future. You can read about it here: https://portfoliocharts.com/2017/03/21/how-to-predict-withdrawal-rates-without-a-crystal-ball/

I had wondered about that and now I find out you wrote about it over two years ago! 

[vand]

It's interesting but it looks like some of the portfolios with a higher SWR have lower returns and vice versa. This is why I'm skeptical of the portfolios that are heavy in gold and cash type assets (bonds, treasuries, etc). Sure we might end up with poor returns in equities markets that result in those portfolios doing well that have gold and cash type assets but there is no guarantee that this will occur.

I like simple and I suppose naive portfolios foot the bill there.

It will be interesting how this changes over the next 50 odd years.

True. So much of this is history-specific. Since 1970, the US went off the gold standard, had a tech boom in equities, inflated real estate prices with government subsidized financing, and from the early 1980s to the present has had a bond rally of massive proportions. Meanwhile investors endured stagflation in the 70s, a tech bubble that burst in 2000, a real estate bubble that burst in 2008, and the worst recession since the 1930s. Will any of that happen in the next 50 years?

Selecting the optimal AA now would require knowledge of the future, not the past. Will there be a biotech boom driven by gene editing? Will bonds revert to their norms or go negative like in Europe? Will Fannie and Freddie Mac be broken up and sold off? Will the tech-driven attention economy hit a natural growth limit or see a backlash as their products become recognized as addictive? Will average savings rates continue to decline or are millennials naturally more frugal? Will there be a major war? Will blockchain disrupt banking and tank the dollar? How will demographic graying affect aggregate demand? Will the US remain a democracy? Will self-driving cars that we don’t own become the new transportation economy? Will parts of New Orleans, Miami, Charleston, Tampa, NYC, and Houston be abandoned and who pays for that? Will China continue growing, stagnate at some point, or collapse? What will be the average inflation rate for the next 3 decades?

If one knows any of this with certainty, portfolio selection becomes easy.

This is exactly how I feel about portfolio theory. It's a guide. You can't predict the portfolios that will be successful in the future because you don't know what will happen in the future.

The purpose of MPT is not to try to predict which assets will be the ones to hold going forward. The point is to help determine how to trade return for risk in the most efficient manner possible.

People who say that they don't see how a multi-asset portfolio can hold up going forward "given the current price of X/Y/Z" should study the components of the Permanent Portfolio:

- A quarter is made up of gold, which got crushed for 20 years between 1980-2000
- another quarter from cash which has been dismal for almost just as long
- another quarter from stocks which had a lost decade from 2010
- another quarter in long term bonds, which people have been trying to call the bottom on for most of the last 20 years

and yet the portfolio's 40 year record still holds up and the strategy continues to perform well in the current FY to date. You don't need everything in a portfolio going up in order for the strategy to be effective.. in fact, the strategy relies on different parts of the portfolio not doing well for periods of time. If every part of your portfolio is doing well,...it's probably not diversified enough.

[ChpBstrd]

Before I lay down a six-figure sum buying an "investment" from someone else (who has reasons they would rather have the cash), I would like to at least have a rationale for why it will earn money or appreciate over time. For example:

"This diversified stock portfolio is expected to return 6% over a decade because earnings will grow with inflation and the economy, because of stock buybacks, because the companies can make leveraged investments and earn a higher ROI than their interest, and because this has been the pattern for decades."

"This rental property is expected to yield an increasing return of 6% to 12% over a decade because [pulls out spreadsheet] expenses will be this much, rents will be this much, and appreciation will be this much assuming a vacancy rate of 15%, rent inflation of 3%, and property and property tax appreciation of 2%."

"This bond portfolio is expected to have a nominal YTM of 4% assuming a coupon of 5% and defaults/write-offs of 1% per year."

What exactly am I supposed to say about gold, or potentially worse, cash?

The best I can think of is:

"I will start by selling monthly at-the-money puts for GLD shares worth half my allocation until I am assigned. Upon assignment of half my allocation, I will sell at-the-money puts for the other half of my allocation, while writing at-the-money covered calls for the shares that have been assigned to me. If my 2nd set of puts are assigned, I will write far-ITM calls for half my allocation until assigned, and then resume the strategy of writing cash-secured puts for half my allocation and covered calls for the other half. If my covered call is assigned, I will write far-ITM puts for half my allocation until assigned, and then resume the strategy. I will make money while doing this because premium received each month will partially offset losses from large moves and because historically the options market's implied volatility has been slightly higher than actual volatility."

...which is pure gambling, but has a stronger rationale than the gambling strategy of buying and holding commodities in the hopes of them appreciating faster than treasuries.

[MustacheAndaHalf]

"Note 4: SWR is determined based on the single worst year in the record for 30-year rolling returns. Average returns would make a very different ranking."

I think average returns are more common, so I'm glad you posted data for that in the replies.

One thing I find confusing: the naive 1st and naive 5th both tilt towards value.  They do quite well.  Larry Swedroe advocates both a value tilt and a small-cap tilt, but didn't do that well.

Maybe it's the small cap tilt, since none of the naive portfolios focus on small cap.

[BicycleB]

Before I lay down a six-figure sum buying an "investment" from someone else (who has reasons they would rather have the cash), I would like to at least have a rationale for why it will earn money or appreciate over time. For example:

"This diversified stock portfolio is expected to return 6% over a decade because earnings will grow with inflation and the economy, because of stock buybacks, because the companies can make leveraged investments and earn a higher ROI than their interest, and because this has been the pattern for decades."

"This rental property is expected to yield an increasing return of 6% to 12% over a decade because [pulls out spreadsheet] expenses will be this much, rents will be this much, and appreciation will be this much assuming a vacancy rate of 15%, rent inflation of 3%, and property and property tax appreciation of 2%."

"This bond portfolio is expected to have a nominal YTM of 4% assuming a coupon of 5% and defaults/write-offs of 1% per year."

What exactly am I supposed to say about gold, or potentially worse, cash?

The best I can think of is:

"I will start by selling monthly at-the-money puts for GLD shares worth half my allocation until I am assigned. Upon assignment of half my allocation, I will sell at-the-money puts for the other half of my allocation, while writing at-the-money covered calls for the shares that have been assigned to me. If my 2nd set of puts are assigned, I will write far-ITM calls for half my allocation until assigned, and then resume the strategy of writing cash-secured puts for half my allocation and covered calls for the other half. If my covered call is assigned, I will write far-ITM puts for half my allocation until assigned, and then resume the strategy. I will make money while doing this because premium received each month will partially offset losses from large moves and because historically the options market's implied volatility has been slightly higher than actual volatility."

...which is pure gambling, but has a stronger rationale than the gambling strategy of buying and holding commodities in the hopes of them appreciating faster than treasuries.

I don't understand either the rationale of why the naive portfolios would often have cash and gold, yet do so well. But I would like my investing rationale to adequately explain the past before I would expect it to predict the future. Even in the average returns table, it looks like the naive portfolios as a group beat the expert ones.

In the period covered by the data, was the value of the uncorrelated variances so large that it produced more return than was lost by the underlying disadvantage of the low-return asset classes?

If so, is it reasonable to think that the value of the uncorrelated variances would continue? That's an investing rationale in itself, right?

[arebelspy]

Man,what a great thread. Thanks for doing all that work and sharing with us, @Radagast !

I was a TSM guy for a long long time, but reading the Permanent Portfolio blog posts and playing with the data eventually changed my mind (it still took years of doing that). Now I happily slice and dice a bit, confident my portfolio is all the better for it.

[AdrianC]

People who say that they don't see how a multi-asset portfolio can hold up going forward "given the current price of X/Y/Z" should study the components of the Permanent Portfolio:

- A quarter is made up of gold, which got crushed for 20 years between 1980-2000
- another quarter from cash which has been dismal for almost just as long
- another quarter from stocks which had a lost decade from 2010
- another quarter in long term bonds, which people have been trying to call the bottom on for most of the last 20 years

and yet the portfolio's 40 year record still holds up...

1980-2019, $10K initial, annual rebalance.
Final Balance:
Permanent Portfolio $172,324
60/40 (Intermediate Treasuries) $456,621

https://www.portfoliovisualizer.com/backtest-asset-class-allocation

Does anyone remember that old Little Caesar's commercial "One banana or two bananas... he seems to prefer two"?
https://www.youtube.com/watch?v=J47o4YZ_yrM&list=PLXUmNlnWp46X18xtT4wjIXjcfTfYuLTb-&index=35

(Note that I'm the orangutan here).

[bacchi]

0th Naive in the USA-Perpetual column isn't ordered correctly.

[bacchi]

There's a lot to unpack here.

If you weren't in the US, buying "international" in the 1970-90s meant buying US. That could explain the good showing of the Ivy portfolio for non-US based portfolios, which has 20% in international stocks and 20% in international bonds (i.e., "international" is majority US for many of the time periods).

F&F might explain why some of the value portfolios do well in other countries. The 3rd does ok to great in most countries, where "international" means US for most of the time periods.

Great chart, Radagast!

[Radagast]

It's interesting but it looks like some of the portfolios with a higher SWR have lower returns and vice versa. This is why I'm skeptical of the portfolios that are heavy in gold and cash type assets (bonds, treasuries, etc). Sure we might end up with poor returns in equities markets that result in those portfolios doing well that have gold and cash type assets but there is no guarantee that this will occur.

I like simple and I suppose naive portfolios foot the bill there.

It will be interesting how this changes over the next 50 odd years.

True. So much of this is history-specific. Since 1970, the US went off the gold standard, had a tech boom in equities, inflated real estate prices with government subsidized financing, and from the early 1980s to the present has had a bond rally of massive proportions. Meanwhile investors endured stagflation in the 70s, a tech bubble that burst in 2000, a real estate bubble that burst in 2008, and the worst recession since the 1930s. Will any of that happen in the next 50 years?

Selecting the optimal AA now would require knowledge of the future, not the past. Will there be a biotech boom driven by gene editing? Will bonds revert to their norms or go negative like in Europe? Will Fannie and Freddie Mac be broken up and sold off? Will the tech-driven attention economy hit a natural growth limit or see a backlash as their products become recognized as addictive? Will average savings rates continue to decline or are millennials naturally more frugal? Will there be a major war? Will blockchain disrupt banking and tank the dollar? How will demographic graying affect aggregate demand? Will the US remain a democracy? Will self-driving cars that we don’t own become the new transportation economy? Will parts of New Orleans, Miami, Charleston, Tampa, NYC, and Houston be abandoned and who pays for that? Will China continue growing, stagnate at some point, or collapse? What will be the average inflation rate for the next 3 decades?

If one knows any of this with certainty, portfolio selection becomes easy.

This is exactly how I feel about portfolio theory. It's a guide. You can't predict the portfolios that will be successful in the future because you don't know what will happen in the future.

The purpose of MPT is not to try to predict which assets will be the ones to hold going forward. The point is to help determine how to trade return for risk in the most efficient manner possible.

People who say that they don't see how a multi-asset portfolio can hold up going forward "given the current price of X/Y/Z" should study the components of the Permanent Portfolio:

- A quarter is made up of gold, which got crushed for 20 years between 1980-2000
- another quarter from cash which has been dismal for almost just as long
- another quarter from stocks which had a lost decade from 2010
- another quarter in long term bonds, which people have been trying to call the bottom on for most of the last 20 years

and yet the portfolio's 40 year record still holds up and the strategy continues to perform well in the current FY to date. You don't need everything in a portfolio going up in order for the strategy to be effective.. in fact, the strategy relies on different parts of the portfolio not doing well for periods of time. If every part of your portfolio is doing well,...it's probably not diversified enough.

First, it figures you would say that, as in the past the Permanent Portfolio did better in the UK than anywhere else. You are probably a little biased though ;)
Second, I can't recommend the portfolios that ended up in red with white text for either of the two charts I posted. Well, I could recommend the No-Brainer, with tweaks.

[Radagast]

"Note 4: SWR is determined based on the single worst year in the record for 30-year rolling returns. Average returns would make a very different ranking."

I think average returns are more common, so I'm glad you posted data for that in the replies.

One thing I find confusing: the naive 1st and naive 5th both tilt towards value.  They do quite well.  Larry Swedroe advocates both a value tilt and a small-cap tilt, but didn't do that well.

Maybe it's the small cap tilt, since none of the naive portfolios focus on small cap.

Of course they are more common :D. One thing is that the "average returns" favor portfolios with greater volatility because they use the arithmetic mean. Take a portfolio which returned 20% per year then lost 80% in the tenth year. It is 1.2*1.2*1.2*1.2*1.2*1.2*1.2*1.2*1.2*0.2 versus (0.2+0.2+0.2+0.2+0.2+0.2+0.2+0.2+0.2-0.8)/10 which give very different results. However if you are making monthly contributions and are less than half way to your FIRE goal, then the arithmetic average is exactly what you should be using (yay me!) A person looking at average safe withdrawal rates would be looking somewhere between the two charts posted here.

I think all the portfolios focus on small cap. "Large Blend" and the other large slices are just a little bit larger than total market, but all the portfolios have more mid and small cap slices than large cap. Thus, every portfolio has a small cap tilt, it is a large cap tilt which is missing (because Portfolio Charts does not feature a megacap selection). It is growth and value which do not have a bias one way or another with my/Portfolio Charts/conventional acedemic method, although you can find either in practice.