Dissecting the 60/40

The CFA Institute produced a detailed report on the historical performance of the 60/40 portfolio (here the link). While the breadth of this study is not as ample as other works, think of the UBS Yearbook, I found some interesting graphs…and the hook to write about potential issues in using the mean/variance optimisation.

I do not have the data to corroborate this idea but I think the US “exceptionalism” is driven by the fact that for a large part of the period, US Treasuries were the ‘safe asset’ when a crisis happened. Other parts of the World had idio-crises, many European countries like Germany and Italy, that drove that correlation up. Consider for example how WWI and WWII impacted Europe and Japan compared to the US and Australia.

The bond index is weighted by Gross Domestic Product. By now, you should probably know that there is no great way to build a bond index but GDP-weighted is probably leaning too far away from ‘safe heaven’. I like what NTSG is doing in this regard. I would say that the right expected correlation is between 0.48 for World and 0.19 for the USA.

Not that all of this matters that much, since I am more interested in the tail hedge property of bonds.

A higher correlation shifts each point to the right and bottom, meaning each portfolio is more risky and less profitable.

On the other hand, if you manage to find uncorrelated assets, their ensemble would produce higher returns for the same risk—for free.

These are REAL returns. I am a firm believer that the picture Since 1970 and Since 1990 fucked up a lot of investors and you will be damned if you think it is replicable. Better to work on the assumption that the future looks like the first set of bars.

It does not mean dumping all your bonds. But if you have only stocks and bonds, stocks would be the workhorse and bonds would have limited capacity to save you when shit hits the fan. That’s why you need to be more diversified.

I like this graph because it is immediate in showing how heuristics can deceive you. Instead of insisting on the 60/40 we should probably highlight an “area” of efficiency that goes from the 60/40 up to the 80/20: where you land is then driven by your risk appetite.

Building portfolios of Domestic and International markets brings the curves much closer:

[If you spot why the Global portfolio is performing so poorly compared to all the rest please let me know, it doesn’t make sense to me. The Europe portfolio, for example, should include less good assets (US) and more bad ones (Europe) than the Global portfolio]

Adding Other Asset Classes

The other day I was listening to this episode of the Rational Reminder pod. One of the questions in the episode was related to Return Stacking: Ben described well the discussion that happened in their forum between Corey Hoffstein and Scott Cederburg, the pros and cons of leveraging a stock&bond portfolio compared to a 100% stock solution.

Curiously, the mic-drop moment for Scott (and Ben) was the fact that bonds sucked in the past as long-term assets. This is what you see in this paper: outside of a few cases, the US and Australia, bonds performance was lacklustre. But again, this result is skewed, in my opinion, by the fact that investors should tilt their bond allocation towards the reserve currency-real risk-free asset (see more here).

Even in that case, yes bonds largely suck.

That’s why looking at ‘return stacking’ as a fight between bond+stock or just stocks is like judging a Swiss knife by its usefulness as a steak knife. Sure, you can use it like that but you should not buy it with that in mind. Return Stacking is great because it allows to add more asset classes in the same portfolio without having to sacrifice too much on the return side. It is diversification by addition and not by subtraction.

When equities are in a severe drawdown, in the above picture they show episodes with drawdowns higher than 20%, bonds come to the rescue…but not always. That’s why if you think it will always work, you are in the wrong place. The solution is to complement bonds with other diversifiers.

The basket of diversifiers used in this research paper is a bit silly (and also their weightings: BTC equally weighted with public stocks? sure). I would have expected at least a split between gold and other commodities, if they really wanted to test multiple options before looking at HY bonds (high correlation with equities), private equity (high correlation with equities) and hedge funds (huge dispersion in returns between funds).

The highlighted rows are the only 2 that make sense. EBCIP differs from CGP by switching a 10% allocation from World equities to commodities. The result is what we would expect: all metrics improve except the median return, which is the return only your Excel file will ever achieve 🙂

For whatever reason, in the above picture they decided to plot the EBCIP version that allocates 30% to commodities instead of equities. While this configuration has the best historical Sharpe ratio, I would doubt anyone will ever consider it. What is relevant, also for EBCIP 10%, is that the optimal frontier moves up and to the left.

Diversification works and it works even better the more (real) diversifiers you consider.

If we go back to the Rational Reminder ep I cited before, someone might pick Ben Felix’s rebuttal that CAPM is a monoperiodic optimisation. Here is what Michael Mauboussin wrote in his paper “Probabilities and Payoffs” that I discussed here.

“One way to think about this is that mean/variance optimization (Markowitz) focuses on diversification at a point in time and geometric mean maximization (Kelly) considers diversification over time. Markowitz was fully aware of Kelly and related research and wrote favorably about it.”

“Capital accumulation is a multiplicative process, which means that understanding geometric averages, risk management, and portfolio construction are all essential for compounding wealth.“

Michael was a guest on the RR podcast just 1 episode before that one…

The Markowitz framework hits a snag when you’re looking at long horizons and stock returns aren’t (allegedly) IID. Stock index returns show mean-reverting characteristics over the long run (assuming they don’t hit zero).

To prove stock returns aren’t IID, you need more data than we have. Cederburg’s approach of bootstrapping 10-year blocks leaves him with just a few independent blocks, which is why he included many international markets. But is the Austrian stock market in 1900 really representative of today’s globally diversified indexes?

Whether IID or not, the elephant in the room is that stock returns can underperform or overperform for extended periods. If you can genuinely stick to a 30-year horizon, stocks seem less risky than their annual volatility suggests, meaning a portfolio can hold more stocks than CAPM implies.

However, investors must endure those frequent 40%+ drawdowns and still sleep at night. Considering behavioural aspects and the possibility of needing money earlier than expected, those 30-year blocks should be viewed as a series of shorter intervals because that’s how we live.

Cederburg’s optimal 100% stock allocation in some scenarios goes through a 90% drawdown. He points out it happened in 0.00X% of cases. My point is, you need to consider all scenarios where stocks lost more than 65%. The breaking point comes much earlier because we don’t know the future. We can’t be sure that stocks will bounce back this time.

This echoes concerns from folks on this side of the pond about US stock valuations. If stock returns aren’t IID, something’s got to give, and US stocks must come down. So, they sought shelter elsewhere, opting for non-US stocks.

Investing 100% market-cap weighted and believing in non-IID while entering retirement (at some point, you have to consume those savings) means either trying to time the market (knowing sub-par returns are coming) or hoping the sequence of return risk doesn’t blow you out (again, due to those sub-par future returns).

Or you can diversify differently: invest in other asset classes 😉 Knowing this will work better for horizons less than 30 years.

Pick your fighter.

What I am reading now:

Follow me on Bluesky @nprotasoni.bsky.social