Ask whether gold or stocks perform better long term and you will hear two confident, contradictory answers, both backed by real data. Here is the honest version up front. Over the last 40 years, stocks win clearly: from 1985 to 2025, the S&P 500 compounded at roughly 11.9% per year while gold returned about 6.7%, according to research published by JPMorgan. Over the last 20 years, gold actually edged stocks: from 2005 to 2025, gold compounded at roughly 11.6% per year against about 10.7% for the S&P 500, per the same JPMorgan analysis.
Both statements are true. Neither is a trick. The entire difference is the starting point. The 40-year window begins after gold's brutal early-1980s collapse and captures two of the greatest equity bull markets in history. The 20-year window begins just before the global financial crisis and includes gold's two strongest decades since the 1970s. Whoever picks the window wins the argument, which is exactly why the argument itself is the wrong frame.
The argument for equities is structural, not sentimental. A share of stock is a claim on a productive business: it generates earnings, those earnings get reinvested or paid out as dividends, and the reinvested portion compounds. Gold does none of that. An ounce of gold in 1985 is the same ounce today. It pays nothing, builds nothing, and earns nothing. Its return comes entirely from what the next buyer will pay.
Over long horizons, that structural difference is enormous. At 11.9% per year, $10,000 invested in 1985 grows to roughly $900,000 over 40 years before taxes and costs. At gold's 6.7%, the same $10,000 grows to roughly $134,000. That is not a rounding error. It is the compounding gap between an asset that produces cash flows and one that does not, and it is why the default long-term core of most portfolios is, and should remain, equity exposure.
So why does anyone hold gold at all? Because equity returns do not arrive in a smooth 11.9% line. They arrive in regimes, and in some regimes stocks go nowhere for a very long time while gold does the compounding.
The 2000s are the canonical example. From 2000 through 2009, the S&P 500 lost money over the full decade, bracketed by the dot-com crash and the global financial crisis. Gold roughly tripled over the same stretch. An investor who retired into that decade holding only equities lived a completely different financial life than the 40-year averages suggest.
The pattern repeated in miniature in 2022, when the S&P 500 drew down about 25.4% peak to trough while gold essentially held its ground. And it has been running at full volume recently: gold returned about 27% in 2024 (Statista) and roughly 65% in 2025, its best calendar year since 1979 (Visual Capitalist), driven by forces we break down in what actually drives gold prices. That surge is precisely what pulled the 20-year comparison in gold's favor, and it is also central to where the major banks think gold goes from here.
The buyer base behind those numbers has also changed. Central banks purchased 863 tonnes of gold in 2025 per the World Gold Council, and a record 45% of central banks in the WGC's 2026 survey said they plan to increase their gold reserves. That is a structurally different demand picture than the one that prevailed through most of the 40-year window, and it helps explain why the recent regime has looked so unlike the long-run average.
A common assumption is that gold is the wild, speculative asset and stocks are the respectable one. The volatility data does not support that. Across long measurement windows, gold's annualized volatility has generally run in the 10-18% range, which is comparable to broad equity indexes, not meaningfully wilder. In 2022, it was the equity index that fell 25.4%, not the metal.
Gold's real risk is different in kind: it is regime risk, the possibility of long stretches where the price simply goes nowhere. More on that below, because it cuts against gold's current popularity.
The strongest finding in the research is not about gold or stocks in isolation. It is about the combination. Gold's correlation to equities is low and unstable, which means it frequently zigs when stocks zag. The World Gold Council's portfolio research found that adding gold in the range of 4-15% to a diversified portfolio historically improved risk-adjusted returns, meaning more return per unit of volatility than either a pure equity book or a heavy gold book delivered on its own.
That diversification job used to belong to bonds, but bonds failed it badly in 2022, when stocks and bonds fell together for 14 consecutive months. That failure is a large part of why strategists have been rethinking the classic stock-bond mix, a debate we cover in is the 60/40 portfolio dead. Gold held its ground through exactly the environment that broke the old playbook, which is why it has been pulled back into the allocation conversation.
The gold versus stocks debate is usually not a debate about assets. It is a debate about which 20-year window the person arguing happened to live through.
Here is the part both camps need to hear. Buying gold today because it just returned 65% in a single year is a recency bet, and gold's history contains a warning that should be stated plainly: from 1980 to 2000, gold went essentially nowhere for 20 years. An entire generation of investors bought the top of the 1980 spike and waited two decades to break even, while equities compounded relentlessly through the same period. Nothing about gold's structure prevents that from happening again.
But recency risk runs the other direction too. Assuming equities always deliver 11.9% because that is what the last 40 years averaged ignores that the same dataset contains a full decade of negative stock returns. The 2000s were not an anomaly that has been engineered away. They were a regime, and regimes repeat.
Once you see the window problem clearly, the versus framing dissolves. The useful question is not which asset wins, because the answer changes with the start date. The useful question is how much of each to hold so that no single regime can wreck the plan.
Mainstream guidance clusters in a 5-15% band for gold, with 10% the most commonly cited figure, against an equity-dominated core. The WGC's 4-15% range brackets the same territory. We walk through the full allocation math, including the more aggressive cases, in how much gold belongs in a portfolio. The point of a band is discipline: it forces trimming after runs like 2025 and adding during dead zones like the 1990s, which is the opposite of what recency instinct tells people to do.
The equity side is largely a solved problem: low-cost index funds capture the market's compounding with minimal friction, and the SPIVA data showing most active managers underperforming over long horizons argues against paying up for stock picking.
The gold side offers three main lanes. Gold ETFs provide liquid, low-friction exposure and have been absorbing record flows, with $89 billion of inflows in 2025 per the World Gold Council. Physical bullion offers direct ownership at the cost of storage, insurance, and wider spreads. And systematic, rules-based strategies attempt to hold gold exposure when trend and momentum conditions favor it and step aside when they do not, an approach aimed at the regime problem itself rather than at a static weight. Each lane carries different costs and different risks, and none of them removes the fundamental fact that gold's return arrives in streaks.
Fifty years of data supports one boring conclusion: stocks for the engine, gold for the shock absorber, sized by rule rather than by headline. The investors who got hurt in this debate were almost never the ones who owned both. They were the ones who went all-in on whichever asset had just finished winning.
Over the last 20 years, yes: JPMorgan research puts gold at roughly 11.6% annualized from 2005 to 2025 versus about 10.7% for the S&P 500. Over the last 40 years, no: from 1985 to 2025 the S&P 500 compounded at roughly 11.9% versus about 6.7% for gold. Both figures are accurate; the answer depends entirely on the measurement window.
Not in the way most people assume. Gold's long-run annualized volatility of roughly 10-18% is comparable to equities. Gold's distinct risk is long stagnant regimes, such as 1980-2000 when it went essentially nowhere for two decades. Its distinct strength is holding value during equity crises, such as 2022 when the S&P 500 fell about 25.4% while gold held its ground. Neither asset is categorically safer; they fail in different environments.
History argues against all-or-nothing switches in either direction. Selling equities after gold's 65% year in 2025 is a recency-driven decision, and gold's 1980-2000 dead zone shows how badly chasing a gold peak can end. Research from the World Gold Council suggests the historical sweet spot was owning both, with gold at roughly 4-15% of the portfolio improving risk-adjusted returns rather than replacing equities outright. This is an educational overview, not advice for any individual situation.
Gold has historically been the stronger crisis asset. During the 2000s, a decade bracketed by two recessions, the S&P 500 lost money while gold roughly tripled. In 2022, stocks drew down about 25.4% while gold was essentially flat. Stocks, however, have historically dominated the recoveries and expansions that follow, which is why the long-run 40-year numbers favor equities.
Mainstream allocation guidance keeps equities as the core and gold as a satellite in a 5-15% band, with 10% the most commonly cited figure. World Gold Council research found 4-15% allocations historically improved risk-adjusted returns. The right number depends on time horizon, income needs, and risk tolerance, which is a conversation for a qualified advisor rather than a rule of thumb.