Worthwhile article and thread on "The Most Efficient Portfolios"

[kevink]

There's an excellent new article on the Portfolio Charts site that should be of special interest to those trying to avoid "left tail" risk while still enjoying decent returns:

https://portfoliocharts.com/2021/12/16/three-secret-ingredients-of-the-most-efficient-portfolios/

There's much of interest in the piece but one of the things that stood out for me is the counterintuitive nature of many of the most successful (in terms of risk-adjusted returns) allocations. It's fascinating to see how including small amounts (~10%) of assets many of us have little or no interest in owning (long-term Treasuries, US small-cap value, gold most of all) has dramatically boosted safe withdrawal rates.

If the original article isn't enough it has spurred a very lively (and epic in length) thread on Bogleheads:

https://www.bogleheads.org/forum/viewtopic.php?f=10&t=364950

 

[OldShooter]

The fundamental idea of using the efficient frontier is that past variance and past returns are predictive of the future. Underlying that is the idea that variance is a parameter that measures risk.

It's a fun kind of backtest though, and makes backtesting look very scientific.

But if we could accurately predict future returns, why not just skip the graph paper and buy what is going to go up?

IMO anyway.

 

[OverThinkMuch]

If you use different start and end dates, you get different efficient portfolios. To me that suggests the approach is flawed, because in 10 years their current best portfolio probably won't come out on top. If the efficient portfolio keeps changing, I don't want to keep changing my holdings. I found "efficient market theory" interesting, but I don't really apply it to my portfolio.

 

[lem1955]

The return on these super-diversified portfolios comes largely from rebalancing. With many asset classes rebalancing is not a simple exercise if one has multiple accounts with different tax statuses, goals, and positions. For more information about efficient portfolios see Asset Allocation-Balancing Financial Risk by Roger C. Gibson.

 

[Montecfo]

I have long been a fan of small cap value and it's nice to see some validation of what I have already found.

But he makes a good point that you have to own things you have confidence in (and hopefully for sound reasons). Otherwise you will lack the ability to stay the course.

At this time of market inflection I think it is particularly important to know why you own what you own, and where we are in the cycle.

Anybody loading up on long-term Treasuries here? Gold?

But the case for small cap value particularly in an inflationary environment is I think rather strong.

 

[kevink]

If you use different start and end dates, you get different efficient portfolios. To me that suggests the approach is flawed, because in 10 years their current best portfolio probably won't come out on top. If the efficient portfolio keeps changing, I don't want to keep changing my holdings. I found "efficient market theory" interesting, but I don't really apply it to my portfolio.

Tyler addresses this early on in the article:

"The 15-Year Baseline Return solves a major problem with most efficient frontier calculations where the results vary wildly by the timeframe studied. It looks at all 15-year timeframes since 1970 simultaneously and reports the 15th-percentile inflation-adjusted CAGR rather than simply the always-shifting numbers over a single period. Basically, it accounts for the return risk that people always talk about but gloss over with an average. It presents a more honest number of something to safely plan around, and it’s my favorite measure of a conservative long-term return using historical data."

The whole point of the piece is to look at worst-case real (i.e. inflation-adjusted) returns and safe withdrawal rates over every time period since 1970. And if you look at the charts featured the same group of portfolios comes out on top over time.

I think Montecfo's point is also well-taken. It's great to look back at the data as the article does but also important to look forward and I doubt many here would be inclined to take a major position in long-term treasuries or for that matter gold. OTOH small-cap value stocks used judiciously to offset the lopsided dominance of the tech giants in Total Stock Market. For that matter, adding as little as 10% each of SCV, LTT's and gold to an all TSM portfolio has historically made a huge difference in both returns and resiliency to drawdowns.

 

[Magus]

My main two critiques of a well thought out article is

1) REITs are not good indicators for Real Estate for a few different reasons
1a) REITs tend to have assets focused on urban core markets rather than the broad US
1b) REITs, like other public equities, get multiple compression when markets sell off, especially since many of them are very highly levered (and until fairly recently, were in the "banks" categories for passive funds so when people wanted to sell off banks they also sold REITs with passive investing
1c) The private transaction RE market is massively larger than the public REIT market, is significantly more stable, and actually has higher returns since no public company costs.

If you layer on median home sale prices for example you'll get nearly 180 degree difference in returns than public REITs. We don't have enough data but I think if you layer in investing yourself or using sites like Fundrise or Crowdstreet to invest in real estate without using REITs, you will likely significantly boost your returns and lower your beta.

2) Past correlations may not hold up to future correlations. Most immediately, what the Federal Reserve for example has done the last 21 months it has never done before in its history by adding $5-6T to the economy in a very short period of time. Who knows the long term ramifications of it? Crypto is also a new asset class and crowdfunding investments are also new. Bonds yields are also at record lows now, US Debt to GDP is at record highs and Fed is comfortable with higher inflation. I'm sure we'll get a few new wrinkles the next 50 years than the last 50.

 

[OldShooter]

"The 15-Year Baseline Return solves a major problem with most efficient frontier calculations where the results vary wildly by the timeframe studied. ... "

Hello, ... hello, ... Earth to Author: It is not a problem that results vary wildly. It is the real world. Even a cursory look at a quilt chart will dispose of the notion that past performance indicates future results. Averaging over 15 years (or 30 years or 50 years) papers over the problem in the backtest but it does not change reality going forward.

On a related topic, it took me a long time to figure out why Markowitz equates variance to risk. Note that variance can be positive or negative in his world. So, what it is measuring is the risk that the chosen asset will vary from its predicted return. Positive or negative is the same -- a miss; performance outside the bounds of the efficient frontier prediction.

Sharpe and Sortino have tried to fix this by computing numbers that look only at the "down" variances, but the fundamental problem remains. Variance is not risk. GE, Sears, Enron, Montgomery Ward, Worldcomm, JDS Uniphase, Theranos, JCPenny, ... the list of stocks that showed us real risk is endless.

 

[captain3d]

Interesting article. The more I read the more I thought it was heading towards the Permanent Portfolio that I follow for my core holdings and sure enough it popped up at the end.

All the benefits of the diversification come at key stress points where some thing crashes and rebalancing is triggered

 

[Montecfo]

My main two critiques of a well thought out article is

1) REITs are not good indicators for Real Estate for a few different reasons
1a) REITs tend to have assets focused on urban core markets rather than the broad US
1b) REITs, like other public equities, get multiple compression when markets sell off, especially since many of them are very highly levered (and until fairly recently, were in the "banks" categories for passive funds so when people wanted to sell off banks they also sold REITs with passive investing
1c) The private transaction RE market is massively larger than the public REIT market, is significantly more stable, and actually has higher returns since no public company costs.

If you layer on median home sale prices for example you'll get nearly 180 degree difference in returns than public REITs. We don't have enough data but I think if you layer in investing yourself or using sites like Fundrise or Crowdstreet to invest in real estate without using REITs, you will likely significantly boost your returns and lower your beta.

2) Past correlations may not hold up to future correlations. Most immediately, what the Federal Reserve for example has done the last 21 months it has never done before in its history by adding $5-6T to the economy in a very short period of time. Who knows the long term ramifications of it? Crypto is also a new asset class and crowdfunding investments are also new. Bonds yields are also at record lows now, US Debt to GDP is at record highs and Fed is comfortable with higher inflation. I'm sure we'll get a few new wrinkles the next 50 years than the last 50.

Good points. You know, everyone's investment approach (and tools like Firecalc) focus on past performance. That's understandable: no one has a crystal ball.

But the lack of a crystal ball is a hurdle to all investment strategy discussions.

Analysis is really all we have and the article is notable in the analysis.

 

[chassis]

There's an excellent new article on the Portfolio Charts site that should be of special interest to those trying to avoid "left tail" risk while still enjoying decent returns:

https://portfoliocharts.com/2021/12/16/three-secret-ingredients-of-the-most-efficient-portfolios/

There's much of interest in the piece but one of the things that stood out for me is the counterintuitive nature of many of the most successful (in terms of risk-adjusted returns) allocations. It's fascinating to see how including small amounts (~10%) of assets many of us have little or no interest in owning (long-term Treasuries, US small-cap value, gold most of all) has dramatically boosted safe withdrawal rates.

If the original article isn't enough it has spurred a very lively (and epic in length) thread on Bogleheads:

https://www.bogleheads.org/forum/viewtopic.php?f=10&t=364950

Thanks for the thoughtful post.

I ascribe to efficient frontier as one of many elements of portfolio construction and management.

Cash is a diversifier. Take a look at what a smallish cash position does for portfolio volatility and return. It’s a great tool in the portfolio arsenal, cash.

Also thanks to @Montecfo for the continuous contribution of excellent posts.

To any investor who is frozen in indecision by warnings of “past isn’t a guarantee of the future”:

1. Ignore such worry-wartedness
2. Learn how to assess, choose and manage investments. Anyone can do it. Please don’t believe me, I’m just a guy on the internet. Do believe Peter Lynch. He says the same thing. He’s a billionaire.

 

[Magus]

Good points. You know, everyone's investment approach (and tools like Firecalc) focus on past performance. That's understandable: no one has a crystal ball.

But the lack of a crystal ball is a hurdle to all investment strategy discussions.

Analysis is really all we have and the article is notable in the analysis.

Agreed, but I think its always worth keeping in mind as left tail risks in particular that folks worry about will likely not be the exact same over the next 50 years as the past 50 years.