The job description bonds stopped fulfilling
The forty in a balanced portfolio has one contractual obligation: to rise, or at least to hold, when equities fall. For most of the four decades after 1982 it met that obligation handsomely, because a disinflationary world let central banks cut rates into every growth scare, and falling rates lifted bond prices precisely when equities were under pressure. The negative stock-bond correlation that made 60/40 elegant was not a law of finance. It was a feature of a particular regime, and the regime has changed.
In 2022 the change announced itself. A global aggregate bond index fell by roughly 16 per cent in dollar terms, its worst calendar year in the history of the modern indices, in the same twelve months that global equities fell by close to 18 per cent. The ballast did not merely fail to offset the equity decline; it amplified it. When the shock is inflation rather than growth, the duration that usually protects becomes a second source of loss, because the same higher rates that compress equity valuations also mark down the bonds held to defend against them.
Gold, meanwhile, did the job. Through 2022 it finished broadly flat in dollar terms and meaningfully positive in euro, a near-perfect outcome for an asset whose only mandate in a portfolio is to be uncorrelated when correlation matters most. The pattern repeated through the rate-and-tariff turbulence of 2025. Measured over rolling two-year windows, the correlation between gold and the drawdowns of a conventional 60/40 has fallen to its most negative reading since 1973, the year the dollar's last formal link to gold was severed. That is not a coincidence of dates so much as a reminder of what gold is for.
Why the protection is structural, not sentimental
It is tempting to dismiss a strong run in gold as a fear trade that will mean-revert. We think that reading mistakes the mechanism. Gold's defensive value in the present regime rests on two properties that are structural rather than emotional. The first is what it does not contain: gold has no duration, no coupon and no issuer, so the single risk that detonated the bond sleeve in 2022, sensitivity to the level of real rates working through a long stream of fixed cash flows, simply does not apply to it. An asset with no promised cash flow cannot have those cash flows repriced.
The second property is a genuine change in the buyer base. Since 2022 central banks have purchased over 1,000 tonnes of gold in each calendar year, roughly double the pace of the preceding decade and equivalent to around a quarter of annual mine supply. This is price-insensitive, strategically motivated demand, concentrated among reserve managers in Asia and the emerging world who watched foreign-exchange reserves become a sanctionable asset in 2022 and drew the obvious conclusion. A bar of gold in a domestic vault carries no counterparty and answers to no foreign clearing system.
These two forces interact. Official-sector accumulation has put a higher floor under the price and reduced gold's historical sensitivity to US real yields, the relationship that once made it look like a leveraged bet against the Federal Reserve. The metal has, in effect, been partially de-linked from the variable that used to dominate it, which is exactly why it could rise while real rates were positive. For an allocator, the relevant point is that the diversification on offer is being underwritten by a class of buyer with motives entirely unrelated to the private portfolios it happens to protect.
What the figures actually say
We are wary of arguments that rest on a single flattering window, so it is worth being precise about what the long record supports and what it does not. Over multi-decade horizons gold's real return is close to its function as a store of value: positive but modest, well below equities, with no income to compound. Anyone who buys it expecting equity-like returns has misunderstood the instrument and will eventually be disappointed.
The case is not about return; it is about the shape of the return path. The statistic that matters for a balanced portfolio is conditional behaviour: how an asset performs in the specific months when the rest of the portfolio is falling. On that measure gold has, across both the 2022 and 2025 episodes, contributed positive returns during equity-and-bond drawdowns, the rare combination that the bond sleeve was supposed to deliver and did not. A modest sleeve of an asset that pays off in the left tail can lift a portfolio's risk-adjusted return even when its standalone return is unremarkable, because diversification is paid in covariance, not in headline performance.
None of this requires forecasting the gold price, and we are not doing so. The argument holds whether the metal trades higher or lower from here, because the allocation is justified by its behaviour relative to the rest of the book, not by a view on its level. A position one rebalances into on strength and out of into weakness does its work regardless of where the price ultimately settles.
The counter-argument, taken seriously
The objection to gold is real and deserves to be stated at its strongest. It yields nothing. In a world where high-grade euro and dollar bonds again offer real coupons, holding an asset with no cash flow carries an opportunity cost that compounds quietly year after year. Gold is also volatile in its own right, with annualised swings closer to equities than to bonds, so it is not a substitute for the stability that genuinely low-risk instruments provide.
There is a further, fairer point: the negative correlation we are celebrating is itself regime-dependent. Gold protected against the inflation shock of 2022 because that shock damaged both equities and bonds. It offers far less in a conventional, growth-led recession, the kind in which disinflation lets central banks cut and high-grade bonds rally hard. In that world the traditional bond sleeve remains the superior hedge, and gold may simply tread water. An allocator who replaces bonds with gold rather than supplementing them is trading one regime-specific insurance for another.
We accept these points in full, and they shape the recommendation rather than undermining it. The conclusion is not that gold should displace duration. It is that a portfolio holding only duration is insured against the recession regime and exposed in the inflation regime, and that a small, permanent gold position closes precisely that gap. The two hedges are complements, not rivals, because they pay off in different states of the world.
Sizing and implementation
We re-state our long-standing preference for a strategic gold allocation of 3 to 5 per cent of total assets, held as a permanent feature of the policy portfolio rather than a tactical position to be timed. Below roughly 3 per cent the sleeve is too small to alter portfolio behaviour in a drawdown; much above 5 per cent and the standalone volatility and the drag from zero yield begin to assert themselves on the long-run return. The band is wide enough to flex with a family's circumstances and narrow enough to remain genuinely diversifying.
The funding decision matters more than most clients expect. Because gold is intended to repair the defensive sleeve, we prefer to fund the position chiefly from fixed income rather than from equities, accepting a small reduction in the bond allocation in exchange for protection against the regime in which bonds fail. Funded this way, the addition lowers the portfolio's reliance on a single hedge without raising its equity beta, which is the opposite of what a fear trade would do.
On vehicle, physical and physically backed exposure is what does the diversifying work; gold mining equities, by contrast, behave as a leveraged, equity-correlated proxy and reintroduce the very correlation the allocation is meant to remove. For euro-based clients the currency lens applies here too: gold is priced in dollars, so an unhedged position carries a dollar exposure, though in practice the metal's own moves dominate and a deliberate currency view can be expressed separately. Allocated, segregated holdings are preferable to unallocated claims where the position is sizeable, for the same counterparty reasons that motivate the official sector.
What this means for clients
First, treat gold as part of the architecture, not as a market call. A 3 to 5 per cent allocation belongs in the strategic asset allocation alongside equities and bonds, with a stated rebalancing rule, so that the position is trimmed on strength and topped up on weakness automatically. The discipline of rebalancing is where much of the benefit is captured, and it removes the temptation to chase the price after it has already moved.
Second, fund it from the defensive sleeve and hold it for the long term. The point of the allocation is to make the conservative part of the portfolio robust to more than one kind of shock. Clients who already run a sizeable high-grade bond book should view a small gold position as completing that book rather than competing with it, and should expect to hold it through periods when it appears to do nothing.
Third, hold it in the right form. Physically backed, low-cost exposure in allocated or segregated form delivers the diversification cleanly; mining shares and structured products dilute it. For clients with concentrated wealth or cross-border concerns, the absence of counterparty and jurisdictional risk is itself part of the appeal, and is best preserved by owning the metal directly. The desk can structure the holding and the rebalancing framework around each family's existing allocation and currency base.
This note is house research and reflects the views of the Multi-Asset desk at the time of writing. It is not investment advice or an offer.
