The box is the S&P 500 itself: VOO, SPY, and the funds that track it. It has been the default core holding for most portfolios, and for good reason.
But 26 years of annual return data show that when the S&P 500 has a down year, a small handful of assets have consistently held up better: gold, energy, utilities, and long treasuries. This post looks at two of those, gold and oil, and at the idea of carving out roughly 10 percent of a portfolio for them, not as a replacement for the S&P 500, but as ballast for the years it slumps.
What Is "The Box," and Why Look Outside It?
VOO and SPY are, functionally, a bet on the same 500 large companies weighted by market cap. That is not a criticism: it has been an excellent long-run bet. But it is also a concentrated one. When the index has a bad year, nearly everything inside the box tends to move together, since sector weights inside the S&P 500 are correlated with the broader market cycle.
The point of stepping outside the box isn't to abandon it. It's to hold a small slice of something that doesn't move in lockstep with it, so that a bad year for the S&P isn't automatically a bad year for the whole portfolio.
What Actually Holds Up When the S&P Slumps?
Looking at annual total returns from 2000 to 2025, a picture emerges. In the 20 years the S&P 500 finished positive, growth-oriented sectors dominated: Technology (XLK) led with 7 top-2 finishes, followed by Gold with 6, then Energy (XLE) and Consumer Discretionary (XLY) with 5 each.
Flip to the 6 negative S&P years (2000, 2001, 2002, 2008, 2018, 2022), and the list narrows sharply. Gold, Utilities (XLU), and long treasuries (TLT) each landed in the top 2 twice. Energy, Consumer Discretionary, Financials, Materials, Health Care, and Consumer Staples each appeared once. Technology, Communication Services, and Real Estate never once ranked top 2 in a down year.
Across the full 26 years, gold's 8 top-2 finishes is the most of any asset tracked, edging out Technology's 7. That is the headline case for gold as a diversifier.
Gold's Catch: Great in Crashes, Weak in Between
Here's where the picture gets more honest. A separate empirical study of the same 25-year window found gold posted a 100 percent hit rate as a top-2 asset in every single negative S&P year studied (2001, 2002, 2008, 2018, 2022). In those years, gold averaged +6.2 percent while the S&P averaged -18.2 percent, a spread of roughly 24 percentage points.
But gold's long-run numbers are lopsided. Its 15-year CAGR from 2011 to 2025 was just 5.8 percent, compared to the S&P's 13.1 percent over the same stretch, a gap of about 730 basis points a year. Nearly all of gold's 25-year outperformance came from a single stretch, 2001 to 2010 (gold: +18.0 percent CAGR versus the S&P's -0.95 percent), driven by the dot-com bust, the financial crisis, and a weak dollar. From 2012 through 2024, gold went 13 straight years without a single top-2 finish.
Gold also pays no dividend. Over 15 years, the S&P 500's reinvested dividends alone contributed an estimated $1.31 per dollar to its terminal value, a structural advantage gold simply cannot replicate. And gold isn't calm: its 25-year annualized volatility runs around 14.4 percent, with a -28.0 percent year in 2013.
The honest summary: gold has been a strong crash hedge, not a strong long-term compounder. That distinction matters for how much of it you'd want to hold, and why.
Why Oil Earns a Spot Too
Energy, tracked here through XLE, tells a different kind of story: two distinct boom cycles rather than one long resilient stretch. The first ran through the mid-2000s commodity run, with XLE posting top-2 finishes in 2004 (+33.9 percent), 2005 (+40.2 percent), and 2007 (+36.9 percent) as crude prices surged. The second came after COVID and the Russia-Ukraine war disrupted global energy supply, sending XLE to +53.3 percent in 2021 and +64.2 percent in 2022, the single largest annual return of any asset tracked across the entire 26-year dataset, in a year the S&P 500 was down double digits.
Unlike gold, oil's strength hasn't reliably clustered around S&P downturns. It showed up in exactly one of the six down years (2022), but when it does show up, the moves have been large. That makes it a higher-variance complement to gold rather than a substitute for it: less consistent as a hedge, but with the potential for outsized years that gold hasn't matched since the early 2000s.
How Might a 10 Percent Allocation Work?
None of this is a suggestion to abandon VOO or SPY as a core holding. The idea is smaller and more conservative: carving out roughly 10 percent of a portfolio, leaving 90 percent in the index, and splitting that 10 percent between gold and oil exposure.
One way to think about the split conceptually: weight gold heavier, since its track record during actual S&P down years is more consistent, and treat oil as the higher-volatility, higher-upside piece. A 6 percent gold and 4 percent oil split, or an even 5 and 5, are both reasonable starting frames depending on how much year-to-year swing you're comfortable with. This is a structural idea, not a recommendation, and the right split depends on your own goals, timeline, and risk tolerance.
KEY TAKEAWAY: Gold has been the most consistent S&P 500 down-year hedge over the last 26 years, but it has lagged badly as a long-term compounder outside of crisis stretches. Oil has been less consistent as a hedge but has produced some of the largest single-year returns on record. A modest, roughly 10 percent combined allocation, kept outside the core index position, is one way to access both without abandoning the box that has worked well over the long run.
Frequently Asked Questions
A: The point of a satellite allocation is diversification, not replacement. A small slice is enough to cushion a down year in the core index without meaningfully dragging on long-run compounding if the index continues to outperform, which it has over most 15-year stretches in this data.
A: Gold has the more consistent record specifically during S&P down years (a 100 percent top-2 hit rate across the 5 negative years studied). Oil's big years haven't clustered around downturns the same way, and it carries higher volatility, so it behaves more like a swing-for-the-fences complement than a dependable hedge.
A: No. Both data sets studied here span a specific 26-year window, and gold's outperformance in particular was concentrated in one decade (2001-2010). Past performance in this kind of asset-return data does not guarantee how the next 10 or 20 years will play out.