Building and Risk-Managing Multi-Asset Portfolios with ETFs

When investors ask me about portfolio construction, most of the time they’re looking to maximize their return. However, with my experience in building ETF portfolios, I would argue that this is the wrong target to aim for. Instead, the question you should be asking, and the one that can actually get you a lasting wealth, is return on a risk-adjusted basis.

How much volatility must you stomach to achieve the returns you desire? How much could your portfolio fall during a rough year before you start to worry about exiting the bottom of the market? These are far from the same as the query, “what can I earn,” but they can help guide you to quite distinct portfolios.

A multi-asset ETF portfolio of stocks, fixed income, and commodities is among the purest forms of tackling this issue. Each piece has its own risk profile; when put together, they offset one another. As such, the total is less volatile than the individual parts. Of course, you can’t simply throw a few ETFs into an investor portfolio and call it diversified. Rather, you should have a strategy for selecting what you hold, weighting it, and rebalancing according to market conditions. This requires an understanding of how risk is measured, which starts with two concepts that people generally misinterpret: correlation and beta.

Understanding Correlation and Beta

Correlation is a statistical tool that helps quantify the degree to which two assets move together. If the correlation between two assets is one, they will move in perfect tandem. For example, when asset A goes up by 10%, so does asset B. If the correlation is zero, neither asset moves as a result of the other. And if the correlation is negative one, then when asset A goes up, asset B is going down by that exact amount.

That last category is the holy grail for portfolio construction, but in practice truly uncorrelated assets are rare and constantly shifting. One lesson that I learned the hard way is that correlations have a tendency to converge to one during a panic. When liquidity evaporates, a variety of assets that looked very uncorrelated in calm markets will all drop together. This is what occurred during the 2008 financial crisis, what happened again in March 2020, and what will happen again in the future.

The easiest way to think about how an individual stock moves in relation to the overall market is the following chart:

Typical long-term correlations between major asset classes. Illustrative values based on historical patterns; real correlations vary by period.

Beta is a measure of the degree to which an individual stock moves relative to the general market. Essentially, beta is the slope of the line; the steeper the line, the more the stock moves in relation to overall market movement.

Tech stocks serve as an example here, given that they typically have a beta greater than that of the overall stock market — call it 1.8 relative to the S&P 500 index (with beta of 1). So when the broader market goes up by 10%, a technology ETF has a beta that moves by about 18%, but it also drops by about 18% when the market is down by 10%. Higher beta works both ways.

At the other end are low-beta sectors. A stock with a beta of 0.5 is still following the market direction but with less volatility — for instance, if the market goes up 10%, the stock is going up about 5%. Less growth, less risk. Utilities and oil majors are traditional low-beta stocks.

A healthy portfolio absolutely needs these lower-beta assets. If your portfolio is all tech stocks, you’ll make a bunch of money in a bull market but get tilted in a correction. But with a handful of utility stocks mixed in, the ride gets a lot smoother.

Finally, the rest of the spectrum:

• Beta of 0 — assets that don’t correlate with the stock market. Cash is a good example. If the market is down 30%, your cash is still worth exactly what it was yesterday. Conversely, if the market goes up massively, your cash will make exactly zero dollars.
• Negative beta (below 0) — assets that tend to move opposite the stock market. Long-term Treasury bonds typically fit here, because when equities sell off, investors flee to the safety of government debt and bond prices rise. Gold will fit in this category sometimes. It can be a safe haven, going up when equity markets struggle.

I do need to emphasize this caveat: having 0 or a negative beta does NOT mean your position is free of risk. Cash has inflation risk. Treasury bonds have duration and interest-rate risk. Gold has its own unique risks. Beta only gives you information about how that particular asset correlates to ONE index, one benchmark. It says nothing about the asset’s own internal risks.

That said, to build a well-balanced portfolio, you do need to consider individual stock volatility AND how different individual stocks will correlate to one another. And this is where it becomes fun.

Beta-Weighted Position Sizing: Making Risk Equal

One mistake I see all the time (and one that I made for many years of my investing career) is thinking about portfolio construction in dollar terms, not risk terms.

Let’s say you have $100,000 to invest. You allocate $50,000 to a tech ETF, $50,000 to a utility ETF. Looks like a 50/50 portfolio — classic diversification. But it isn’t!

Let’s say the beta of the tech ETF is 1.8, and the beta of the utility ETF is 0.2 (I’m rounding a bit, but these are in the ballpark of where they actually fall). The beta of the tech position is 9× that of the utility. That 50/50 portfolio in risk terms is actually more like 90/10. Your overall returns are going to be dominated almost entirely by what tech does.

The same dollar-weighted 50/50 portfolio is actually 90/10 in risk terms once you account for beta

This is where beta-weighted positioning is going to save you. You don’t think in terms of dollars, you think in terms of risk contribution. If you want both sleeves of your portfolio to make the same contribution of overall beta, you would need to put closer to $90,000 in the utility ETF and only $10,000 in the tech ETF — because that tech position is 9× more risky than utilities per dollar, so you need 9× the amount invested in utilities to balance it out.

Of course, that doesn’t mean that every position should have equal risk. It rarely does. The point of beta-weighted sizing is that you can choose to make your risk exposure intentional, instead of stumbling into it.

If, say, your portfolio beta is supposed to be 1.2 — a slightly more aggressive version of the market, but still far away from the kind of volatility that accompanies tech stocks — and the tech stock ETF has an 1.8 beta (you’re probably not going to want to bet everything on one sector ETF anyway), you can blend $65,000 in the tech ETF with $35,000 in the utilities ETF (0.2 beta). The math:

• 0.65 × 1.8 = 1.17
• 0.35 × 0.2 = 0.07
• 1.17 + 0.07 = 1.24 portfolio beta

Pretty close to your 1.2 target. You now have a portfolio that is slightly more aggressive than the market, but with the safety net of having some utility exposure when the tech stocks do that thing they do where they drop 30% for a month.

This is a pretty simplified example — real portfolios have dozens of positions and calculating beta-weighted sizing becomes harder the more complex the portfolio gets — but the principle still holds: size a position according to how much risk it adds, not the amount of dollars you’re putting in.

An important note about beta as a metric: it measures how an asset moves in relation to a certain benchmark, but it doesn’t cover unsystematic risk — industry blow-ups, supply chain issues, regulatory shocks, any specific risk to an industry or company that could make it drop. A perfectly beta-weighted portfolio that is all-in on a single sector is still vulnerable to that sector’s idiosyncratic risks. Diversifying across sectors, asset classes, and geographies is still necessary no matter how you do your sizing calculations.

Volatility Targeting: How to Control Your Portfolio’s Swings

Where beta weighting focuses on a relative risk against the benchmark, volatility targeting focuses on the magnitude of a portfolio’s swings over the absolute value of that portfolio, regardless of any benchmark.

If you have a portfolio with an annualized volatility of 5%, that means you would expect, over the course of a year, you should have volatility swings of 5% to the up or to the down side. This would be considered fairly conservative and is probably appropriate for those of you who are retired and those who have a lower tolerance for risk. A target volatility of 15-25% would mean much bigger swings in exchange for much higher potential returns.

Going back to our previous example, a portfolio of nothing but tech stocks is going to have a pretty high realized volatility. Adding in the utilities sector pulls down the realized volatility because the utilities ETF is inherently less volatile.

Here’s where it gets tricky. Volatility is not static. Volatility is volatile, sometimes wildly so. We go through cycles of calm markets where realized volatility drops into the single digits, and cycles of chaos where realized volatility triples in a few weeks. I saw one of those volatility regime shifts firsthand in March 2020, when the VIX went from the mid-teens to over 80 in three weeks or so. Any portfolio that was calibrated to any level of volatility was suddenly double or triple its target.

A way of getting through this is volatility targeting, which essentially rebalances itself dynamically:

• When market volatility spikes (say from an expected 15% realized annualized volatility to 30%), you reduce your risky positions and rotate into cash, money market funds, or short-duration Treasury ETFs. This pulls your expected volatility back to the target.
• When market volatility compresses (say to 10% against a 15% target), you reverse course — moving cash and low-vol positions into higher-volatility assets to earn returns consistent with your target risk.

You’re letting the changing market environment serve as a guide to the proportion of an asset to hold, rather than whether to hold it. It is an active management strategy, but a mechanical one. It depends on a measurable metric, not an emotional decision. That is part of what I love about it.

Tail-Risk Hedging with VIX, TLT, and GLD

Let’s take this balanced approach to the next level. Within a multi-asset portfolio, some of the positions will not only provide lower-risk characteristics but will also serve as hedges when the rest of the portfolio is selling off. We call this tail-risk hedging. The goal is to hold positions that make money (or at least retain their value) when the market enters its most severe states. In my experience, three positions do most of this job.

First, long-duration Treasury bonds, which the TLT ETF tracks. It represents Treasuries with 20 years to maturity and is a traditional equity hedge. In a stock panic, long-duration government bonds are generally bought. TLT did well in 2008. It also had a rally in March 2020. This correlation is not perfect — 2022 is a recent example, during which equities and Treasuries both fell as inflation rose — but in most instances, a market panic results in a long Treasury rally.

Second, you might own gold — like the GLD ETF — as a proxy for owning physical gold. Gold will often hold its value (or rise) as equities sell off, but it also will shine during inflation, currency debasement, geopolitical events, and a broader loss of faith in markets. I’ve seen it do the job many times over: 2008, March 2020, the inflation spike of 2022, and the 2024 bank wobble. Gold won’t always rise as equities fall, but it will rise most when you need its services the most.

Third, consider the VIX, which is a measure of the implied volatility on the S&P 500. Often called the “fear index,” the VIX spikes when the S&P 500 sells off in a major drawdown, making it a good crisis hedge. The drawback is that you can’t invest in the VIX directly. The only access point is through an exchange-traded fund that tracks VIX futures. Most of these vehicles use leverage or are inverse products for short-term trading purposes. They’re not good long-term holds, and the contango cost on VIX futures has eaten alive investors who didn’t know what they bought. Use VIX products as tactical trades, not long-term hedges.

When allocating capital to a multi-asset tail-risk hedge, I usually find it wise to allocate 3% to 10% of the portfolio, combined, between TLT and GLD. The hedge is not meant to make you rich but to survive the bad year.

Sample Multi-Asset Portfolio Allocations

When you look at real-world allocations, it all becomes clearer. The classic balanced portfolio sits at a 60/40 split between stocks and bonds. A multi-asset approach would just add some commodity allocation to that:

• 60% stocks
• 30% bonds
• 10% commodities

Going for higher drawdowns for higher long-run returns could look like:

• 75% equities
• 15% bonds
• 10% commodities

This allows you to maintain your growth engine while protecting against a scenario where equities specifically get hit hard.

Probably the best-known multi-asset portfolio is Ray Dalio’s All-Weather Portfolio, with the idea being to be well-diversified enough to perform decently across all four major regimes of the economy (growth, recession, inflation, deflation). It does this by holding mostly uncorrelated assets:

• 40% bonds (usually via long-duration bonds like TLT)
• 30% stocks (usually the S&P 500)
• 7.5% gold
• 7.5% commodities (usually the DBC or some other broad-commodity ETF)
• The remaining 15% is usually some form of medium-duration bonds in the full All-Weather version

The thinking goes that bonds will tend to do well when deflation and recessions hit, commodities are usually a good inflation hedge, and gold usually will do well when the dollar gets debased or in periods of high market turmoil.

The All-Weather framework is one I have seen do just as bad as it does good over the decades. It worked fantastically in 2008, but the last few years it struggled quite a bit when both stocks and bonds were down simultaneously. All-Weather is far from a sure thing and no allocation framework will save you from every event, but I personally like its overall diversification logic better than the majority of frameworks out there.

Harry Browne came up with another more conservative approach back in the 1980s, called the Permanent Portfolio, where you just hold 25% each of:

• Stocks
• Long-term bonds
• Commodities (especially precious metals)
• Cash / Treasuries

This will grow over time at a more modest rate than 60/40 portfolios, but will also have much less realized volatility. The Permanent Portfolio could be a pretty good set-up for investors who want to prioritize downside protection over high long-run returns.

The Bottom Line

A good portfolio doesn’t come from finding the right trends or investing in the right themes. Investors who do well over time are not necessarily the better stock-pickers or forecasters. The ones who do best are the investors with the best risk frameworks.

Understand how much each asset contributes to overall risk. Don’t pick asset weights by dollars — pick asset weights by risk contributions. Pick a target volatility level and rebalance into that target as the market changes. Be sure you are holding a good slice of assets that will do well (or at least perform fine) when the environment goes bad. Remember: no framework will work during every scenario, but if you want to survive across all the different scenarios, that’s better than optimizing for just one.

Two of the questions I always run through any time I’m reviewing a portfolio are:

• Do my risk contributions from each position line up with what my risk targets are for the overall portfolio?
• If we go into a 30% decline quickly over a few months, what’s my plan of action, and what positions are actually going to make me money during the crash?

If you are answering those two questions correctly, you won’t make the worst portfolio construction mistakes. You won’t get all your calls right — nobody does — but you will have a portfolio that will stick around long enough for compounding to work.