What Was the Best Portfolio Over the Last 50 Years?

A few months ago, I analyzed U.S. stock returns over the last century. Today, I'm going to determine the best portfolio over the last 50 years among the following asset classes:

• U.S. Stocks
• International Stocks
• Real Estate Investment Trusts (REITs)
• U.S. Corporate Bonds
• 10-Year Treasury Notes
• 3-month Treasury Bills
• Gold
• Commodities
• U.S. Homes

To do this, I examined the BullionVault annual return data from 1972-2025 (54 years) to determine the optimal/best portfolio over this time period. But before we look at the optimal portfolio, let's get a better understanding of how these various asset classes performed since 1972.

How Have Assets Performed Since 1972?

Since 1972, investors have experienced a wide range of outcomes including bull markets, bear markets, recessions, inflationary periods, and everything in between. As a result of this, asset class returns have varied widely as well. Below is a chart illustrating this for the nine asset classes listed above: There are a few important things to note about these returns:

• Asset with Most Winning Years = Gold
  • Gold had 13 years (out of 54) where it was the asset class with the highest return. This means that Gold was on top roughly 1 in 4 years.
• Asset with Most Losing Years = Commodities/Gold
  • Both Commodities and Gold were tied for the assets with the lowest return in a given year. Each had 11 years where they had the lowest return.
• Best Overall Return = S&P 500
  • $1 invested into the S&P 500 in 1972 would've become $35.90 (in real terms) by 2025. This blows every other asset class out of the water. For comparison, here is what $1 invested in 1972 would've become by the end of 2025 (in real terms) for the next three highest growth asset classes:
    • Gold = $12.89
    • REITs = $12.26
    • International Stocks = $11.27
• Worst Overall Return = Commodities
  • Commodities had a negative real return from 1972-2025, as $1 invested in 1972 would've shrunk to $0.42 by 2025. More importantly, this negative return came with a high dose of volatility. The standard deviation on Commodities was 14% annually while still producing no long-term growth. As a result, this is an asset class I make sure to avoid.
• Most Volatile Asset = Gold
  • Gold's best year was +106% in 1979 and its worst year followed shortly thereafter with -37.6% in 1981. Across all 54 years, the standard deviation was 24.7%, the highest among our asset classes. If you own gold, be prepared for big wins and losses along the way.
• Least Volatile Asset = 3-month Treasury Bills
  • Unsurprisingly, 3-month Treasury bills are the least volatile asset since they are basically the same as cash (or the risk-free rate). However, this lower volatility comes with a cost—lower long-term growth. $1 invested in 3-month T-Bills in 1972 would've only become $1.14 (in real terms) by 2025. This pales in comparison to the S&P 500's overall growth of $35.90 or International Stocks which grew to $11.27 (in real terms) over the same time period.
• Best Inflation-Hedge = Commodities (and Gold if you're lucky)
  • When we break out the data into high inflation (>4% CPI) vs. low inflation (<=4% CPI) periods, Commodities tend to perform the best among all assets. Despite its poor long-term returns, the average annual return for Commodities during periods of high inflation was about 2.9 percentage points higher than during periods of low inflation while the median annual return was basically identical. This wasn't true for most of the other asset classes. For example, during periods of high inflation, the S&P 500 had an average return that was 11 percentage points lower (and a median return that was 17 percentage points lower) than during periods of low inflation. Similarly, International Stocks, REITs, U.S. Corporate Bonds, and 10-Year Treasuries all performed worse during periods of high inflation. Whether we use the average or median, returns were about 8 percentage points lower (if not more) during these periods.
  • Gold is a complicated edge case. Gold's average annual return was 5.6 percentage points higher during periods of high inflation (compared to periods of low inflation). This is the highest among any asset studied. However, Gold's median return was actually 7.7 percentage points lower during periods of high inflation. This demonstrates that Gold's ability to act as an inflation hedge is based upon a few outlier up years (e.g., +106% in 1979, +59% in 1973, +49% in 1974, etc.) that may not repeat again in the future. For example, Gold recently had a +60% year in 2025, but this was 3 years after inflation peaked in 2022. 

My main takeaway from analyzing this data is that no single asset dominates. Some assets have more growth, some have more risk, and some only perform during certain economic regimes. Whether these asset classes will behave similarly in the future is up for debate. Now that we've looked at how different asset classes have performed (in real terms) since 1972, what would have been the optimal mix to get the highest risk-adjusted return?

What's In the Optimal Portfolio?

When it comes to the portfolio with the highest risk-adjusted returns (Sharpe Ratio) from 1972-2025 (i.e., "Optimal Portfolio"), the answer is:

• U.S. Homes = 36%
• S&P 500 = 27%
• Gold = 18%
• 10-Year Treasuries = 16%
• REITs = 3%

Note that 3-month Treasury Bills are absent because they serve as the risk-free rate, so by definition they can't appear in the Optimal Portfolio. The Optimal Portfolio allocation might surprise you, but a core reason for this is that U.S. homes are less volatile (since they are marked-to-market less often) while still having good returns since 1972. As a result, a U.S. home added a decent amount of return to your portfolio without adding much risk. You can see this in the plot below showing the annual return and risk (standard deviation) of the Optimal Portfolio, the efficient frontier, and a handful of other portfolios and individual asset classes: As you can see, the Optimal Portfolio has the highest return per unit of risk. You can't increase your return from this point without also increasing the amount of risk you take. For example, though the S&P 500 had a real average annual return of about 8.3% (or 77% higher than the Optimal Portfolio's 4.7% per year), it also had a standard deviation of 16.8% per year (or 155% higher than the Optimal Portfolio's 6.6% per year). In other words, owning the S&P 500 would get you 1.77x the average annual return of the Optimal Portfolio, but with 2.55x the risk. Note that these are arithmetic means, which is what the efficient frontier is built from. If we were to use geometric means, the return gap would narrow significantly—6.9% per year for the S&P 500 vs. 4.5% per year for the Optimal Portfolio. The return difference would now be 1.54x (versus 1.77x for the arithmetic return), while the risk difference of 2.55x remains the same. The S&P 500 gives up about 1.4 percentage points a year due to volatility while the Optimal Portfolio only gives up 0.2 percentage points. Not having to recover from big losses is a source of return for the Optimal Portfolio. You can visualize this in the plot below which illustrates the annual drawdowns in each portfolio from 1972-2025: Over this time period, the Optimal Portfolio's maximum drawdown was about 13% (in 2022) compared to 48% for the S&P 500 (in 1974). This difference illustrates why portfolio optimization can be so powerful. In theory, you could take the Optimal Portfolio and lever it up (by borrowing) to generate the same return as the S&P 500 while also taking less risk overall. In practice this is harder than it sounds. Levering the Optimal Portfolio 1.77x only works if you can borrow near the risk-free rate, rebalance without friction, and stomach a possible margin call. As a result, most investors are unable to lever their assets so easily. There is one partial exception though—housing. A 30-year fixed mortgage is the most common leverage available to the typical household. More importantly, this leverage can't be margin called. This means that many U.S. households are already running a partially levered version of the Optimal Portfolio without realizing it. Based on my own calculation of the 2022 Survey of Consumer Finances, the median American homeowner has 51% of their net worth in their home (via home equity). The fact that the Optimal Portfolio suggests a 36% allocation to a home is a convenient result that isn't far off many existing homeowner allocations. Of course, some would argue that your home shouldn't be considered as part of your portfolio since housing is a consumption good. I agree with this sentiment, which is why I re-ran the Optimal Portfolio while excluding U.S. housing as an asset class. Without U.S. housing, the Optimal Portfolio from 1972-2025 would have been:

• S&P 500 = 43%
• Gold = 26%
• 10-Year Treasuries = 22%
• REITs = 9%

The 36% exposure to U.S. housing got absorbed by the remaining four asset classes (+16% to S&P 500, +8% to Gold, +6% to 10-Year Treasuries, and +6% to REITs). After this change, the new Optimal Portfolio looks like a more aggressive version of the Permanent Portfolio (25% stocks, 25% gold, 25% bonds, 25% cash). Swap the 25% cash for stocks and some REITs and it's remarkably similar. You can see this more clearly if we add the Permanent Portfolio to the efficient frontier plot from earlier: This isn't an argument for (or against) the Permanent Portfolio or the Optimal Portfolio. Both of these portfolios have been great historically. But history is the past—yet we have to invest in the future. And, unfortunately, the Optimal Portfolio is the ideal solution to the world of yesterday, not the world of tomorrow. This is why the "best" portfolio will always be an illusion.

Why the Best Portfolio is an Illusion

Imagine the world 50 years into the future. Will the Optimal Portfolio over those 50 years (2026-2075) look like the one from the prior 54 years (1972-2025)?  No chance. How do I know? Because the optimal portfolio in 1998 was quite different from the one today. For example, if we were to solve for the Optimal Portfolio from 1972-1998 (half the years in the data), here's what we would find:

• S&P 500 = 68%
• Gold = 24%
• International Stocks = 6%
• U.S. Corporate Bonds = 2%

There's some similarity to the Optimal Portfolio from 1972-2025 in that it holds a sizable S&P 500 and Gold allocation, but that's about it. The 1972-1998 Optimal Portfolio adds International Stocks and U.S. Corporate Bonds, but completely drops 10-Year Treasuries, REITs, and U.S. Homes. Yes, U.S. Homes, which were 36% of the 1972-2025 Optimal Portfolio, are nowhere to be found in the 1972-1998 one! When I say that the Optimal Portfolio of 2026-2075 won't look like the one from 1972-2025, this is what I mean. It's not because I'm bearish on U.S. stocks, skeptical of Gold, or convinced REITs are doomed. It's because I don't know the future. Maybe the U.S. will default on its debt. Maybe Gold will fall out of fashion. Maybe demographic transition will send REITs toward a slow demise. I have no clue and neither does anyone else. This is what makes portfolio optimization such a cruel art form. You get the perfect solution to a problem that no longer exists. The world changes and what you knew about it changes with it. This is why the Optimal Portfolio is an illusion. Because you can never have it going forward, only looking backward. Because the future is like the past...until it isn't. This has convinced me that the actual optimal portfolio is the one that you can stick with no matter what the world throws at you.  Until next time, happy investing and thank you for reading. If you liked this post, consider signing up for my newsletter. This is post 520. Any code I have related to this post can be found here with the same numbering: https://github.com/nmaggiulli/of-dollars-and-data

---