How much gold: what allocation research actually supports

Gold produces nothing. There are no earnings to discount, no coupon, no rent. Any statement that gold is cheap or expensive is a statement about expected real rates, currency debasement or fear, dressed as valuation.

That absence is also the argument for holding it. An asset with no cash flow cannot default, cannot be diluted, and does not depend on a counterparty remaining solvent. It is insurance whose premium is the real return you forgo.

Gold's correlation to developed-market equities has averaged near zero over multi-decade windows, and has turned negative in several severe equity drawdowns while staying broadly uncorrelated in calm periods. That combination — cheap diversification most of the time, positive contribution when it matters — is the entire case.

It is not reliable in every stress episode. In a liquidity crunch everything sells, gold included, because it is the asset a leveraged holder can actually liquidate. The March 2020 and 2008 patterns both show a sharp initial drawdown followed by a strong recovery ahead of equities. Gold protects against a monetary and currency crisis, not against a margin call.

• Long-run equity correlation: approximately zero, unstable episode to episode
• Negative real rates are the most consistent supportive regime
• Liquidity crises hit gold first and reverse fastest

Mean-variance work on 60/40 portfolios repeatedly places the risk-adjusted optimum for gold in a band roughly between 4% and 12%, with the exact figure moving on the assumed equity risk premium and rate path. Below about 3% the position is too small to change portfolio outcomes; above about 15% the zero real yield begins to dominate long-horizon returns.

The band is not precision. It is the range in which gold is large enough to matter and small enough that its lack of yield does not define the portfolio.

Currency concentration matters more than most investors assume: an investor whose income, property and equities are all denominated in one currency has a larger case for gold than a globally diversified one. So does a portfolio heavy in long-duration bonds, which shares gold's rate sensitivity but carries credit and sovereign risk.

Horizon matters in the opposite direction to intuition. Over ten years and longer, cost drag — premium plus custody — compounds against you, so a long horizon demands cheaper access, not a larger allocation.

• Single-currency exposure across income and assets: argues higher
• Already holding inflation-linked sovereigns: argues lower
• Long horizon: argues for cheaper access, not a bigger position
• Leverage anywhere in the portfolio: argues for liquid, sellable form

The diversification benefit only becomes a realised return if you rebalance. An untouched allocation drifts with price and delivers the volatility without the mechanical gain from selling strength and buying weakness.

Annual or threshold rebalancing at a 5% band is sufficient, and both must be costed: with a 3–4% round-trip spread on retail coins, frequent rebalancing destroys more value than it captures. This is where allocated vault holdings with tight spreads have a structural edge over home-stored coins.

If the position exists to be rebalanced and occasionally sold, it needs a tight two-way market: allocated vaulted bars with a published bid. If it exists as a tail-risk reserve you never intend to trade, physical possession and jurisdictional spread matter more than spread.

Most investors need both, in different proportions, and get into trouble by buying the second and expecting it to behave like the first.

The most frequent error is sizing gold against conviction rather than against the portfolio. A 30% position taken because the macro narrative feels compelling is not a diversifier; it is a directional bet on a zero-yield asset, and it will be abandoned at the worst moment when it underperforms equities for three years.

The second error is double-counting. Gold miners, royalty companies and broad commodity funds already carry gold exposure, and equity beta besides. Counting them toward a gold allocation overstates the hedge and understates the equity risk.

• Size against the portfolio, not against conviction
• Miners are equities with gold beta, not a substitute for metal
• Leveraged or derivative exposure does not perform the reserve function
• Home-stored coins with 4% round-trip spreads cannot be rebalanced cheaply

Gold is quoted in dollars, but a European or British holder’s realised return is the dollar gold move plus the currency move. In several past drawdowns, most of the protection a non-dollar investor received came from the dollar rather than the metal.

That is not a reason to hedge it away — for many portfolios the dollar leg is precisely the diversification wanted. It is a reason to look at gold in your own currency when judging whether the allocation did its job, which is why our price pages quote gold in six currencies rather than one.

Put four numbers on one page before you buy: target weight, rebalancing band, the maximum cost you will accept to rebalance, and the circumstances in which you would sell permanently rather than trim. That page is the entire discipline, and it is what stops a reserve asset being liquidated in the middle of the event it was bought for.

Review it annually against the actual holding, the actual custody fee and the actual spread you were quoted — not against the assumptions you made at the start.

• Target weight and the band that triggers action
• Maximum acceptable round-trip cost to rebalance
• Defined conditions for a permanent sale
• An annual review of fees and spreads, not just price