Gold in an optimized portfolio.
A reading of what the long-run portfolio-efficiency research — Ibbotson Associates, the World Gold Council, Bridgewater — actually establishes about a small physical-gold allocation in a long-horizon portfolio. And what it does not.
Sourced. Structural. Not predictive. No allocation advice.
The question gets asked in different ways. The right one is narrower than most of them.
Should I have any gold? Is it a hedge? A diversifier? An inflation play? A safe haven? The framing matters, because the wrong question has no good answer.
The right question, the one the institutional research has actually addressed for sixty years, is narrower: at what allocation, against what alternatives, does adding gold to a diversified portfolio improve the portfolio’s long-run risk-adjusted return? That is a measurable thing. The research bodies most cited on it — Ibbotson Associates, the World Gold Council’s “Gold as a Strategic Asset” series, Bridgewater Associates’ All Weather work — have been answering it with consistent methodology and consistent direction since the 1970s.
This note does not address whether you, individually, should own gold. That is the wrong question for an Office document. The right question is what the published research establishes structurally about gold’s role in a long-horizon portfolio. Allocation decisions are made with your advisor and counsel; they are not made on a website.
The canonical long-run dataset, and two findings from it that matter here.
Roger Ibbotson, finance professor at Yale and founder of Ibbotson Associates, built the canonical long-run dataset for US capital markets. The Stocks, Bonds, Bills, and Inflation (SBBI) Yearbook has tracked the returns of every major asset class back to 1926. It is the dataset most modern portfolio research is benchmarked against.
Two findings from the Ibbotson long-run record are relevant.
First, the long-run real return on gold is positive but modest. Over the SBBI dataset window, gold delivered roughly 4.6% annualized in US-dollar terms, gross of storage — meaningfully above CPI inflation, well below US large-cap equities. That is consistent with gold’s role: not a wealth-compounding engine on the equity model, but an asset that has held real purchasing power across regimes.
Second — and this is the part that matters more for portfolio efficiency — the correlation between gold returns and US equity returns is low and time-varying. Over the SBBI window, the long-run gold-equity correlation runs near zero, with episodic windows of negative correlation during stress events (1973–74, 2000–02, 2008, March 2020). The negative correlation in stress windows is what makes gold useful as a portfolio component — it does its work precisely when the rest of the portfolio is doing badly.
Ibbotson’s own published research on this — the 2010 paper “Strategic Asset Allocation and Commodities,” co-authored with Thomas Idzorek — concluded that a small allocation to commodities (gold among them) improved the efficient frontier of a stock-bond portfolio across the period studied. The allocation that improved the frontier was small. The improvement was real.
The structural argument is the deeper one. The empirical answer lives in Office Note 01.
The Ibbotson long-run dataset above is the structural argument; it covers roughly a century of capital-markets behavior and is the right window for the portfolio-efficiency case. There is also a more recent empirical answer to the most common counter HAR encounters — the assumption that a directly-titled metals layer is a dead-money allocation.
That answer — gold, silver, and the S&P 500 total-return index charted side by side from April 2001 through April 2026 — is the subject of Office Note 01 — Gold and silver vs. the S&P 500. The data block below is the summarized version, with sources and the start-date caveat in writing:
- Gold~+1,735%~12.3% CAGR
- Silver~+1,614%~12.0% CAGR
- S&P 500 — total return~+675%~8.5% CAGR · with dividends reinvested
The window begins near a generational low for both metals — gold around $260/ozt and silver around $4.40/ozt in spring 2001. The result is real; the starting line is favorable. Both facts go in writing.
The Office’s posture, said directly: the structural case for a small directly-titled metals layer rests on the long-run drawdown and correlation evidence above, not on the recent cumulative-return record. The recent record refutes the “dead money” prior. It does not, on its own, justify an allocation decision. The Strategy Brief is the document that reads both windows alongside the household’s portfolio context and writes the conclusion down.
A trade body, read against rather than around, with weight from its methodology.
The World Gold Council is the trade body of the global gold-mining industry, which makes it a self-interested party — a fact the Office reads against, not around. What gives the WGC research weight is methodology: most of the multi-decade portfolio studies the WGC publishes are conducted in partnership with independent academic and institutional researchers (Oxford Economics, New Frontier Advisors, the State Street SPDR research desk) and use the same return data the rest of the institutional research community uses.
The recurring finding across the WGC’s “Gold as a Strategic Asset” series is structurally consistent with Ibbotson: across the post-1970s window — the period of free-floating gold markets — portfolios with a 2%–10% gold allocation have higher risk-adjusted returns than equivalent portfolios without one. The improvement comes principally from drawdown reduction in stress periods, not from outsized gold returns in good periods.
The WGC’s 2024 strategic-allocation update, drawing on 1971–2023 data, places the allocation that maximizes Sharpe ratio for a US-investor multi-asset portfolio in the range of 4%–6%. The exact number varies with the assumption set; the direction does not. Studies that vary the methodology — substitute a different inflation assumption, a different rebalancing cadence, a different set of comparison assets — produce different exact numbers but the same direction.
The conceptual home for all of this is Markowitz’s 1952 framework.
Harry Markowitz published the mean-variance framework — the foundation of modern portfolio theory — in 1952; he was awarded the Nobel Prize for it in 1990. The framework’s core claim: the right way to think about an asset is not in isolation but in the context of the portfolio it joins. An asset that returns less, on its own, can still improve the portfolio’s risk-adjusted return if its correlations to the rest of the holdings are sufficiently low.
That is exactly the structural argument for gold. Gold does not need to outperform equities to belong in a long-horizon portfolio. It needs to do something equities cannot — perform when the equity-bond axis is under stress — at low enough cost (real return drag, storage, transaction friction) that the portfolio overall comes out ahead.
The research establishes that this condition has held empirically over the post-1970s window. The research does not establish that it must hold in the future. That is the genuine epistemic limit of the work, and the Office reads it as such.
Three findings the institutional research bodies converge on.
Across the three research bodies most cited on the question — Ibbotson SBBI, the World Gold Council, Bridgewater — three convergent findings:
- A small gold allocation (typically published in the 2%–10% range) historically improved the risk-adjusted return of a diversified long-horizon portfolio, principally via drawdown reduction in stress windows.
- The improvement is structural in character — gold’s behavior in stress is the mechanism — rather than driven by gold’s standalone return profile.
- The empirical effect is durable across multiple stress regimes (1973–74 stagflation; 2000–02 dot-com; 2008 GFC; March 2020 liquidity seizure; the 2022 bonds-and-equities drawdown), which is what gives the finding institutional weight beyond a single regime’s data.
The part most retail commentary skips.
The research is precise about what it shows. It is also precise about what it does not.
It does not establish a single correct allocation percentage. The 2%–10% range is the range across studies. Inside that range, the right number for any specific household depends on the rest of the portfolio (concentration in equities, exposure to inflation-linked assets, liquidity needs, holding period) — variables that vary household by household.
It does not establish a tactical timing answer. The research is about strategic allocation across multi-decade horizons; it has nothing to say about whether to add gold this month versus next quarter. Studies that have looked at tactical timing of gold allocations report mixed and methodology-dependent results — there is no robust signal in either direction.
It does not establish that gold is the only asset that does this work. Long-duration Treasuries did similar drawdown work in earlier regimes; broad commodity baskets do partial versions of it. What gold has that those alternatives do not is a longer behavioral track record across more regime changes — which is the institutional reason central banks hold gold and not (e.g.) commodity index futures as part of their reserve assets. This is the bridge to Office Note 03.
It does not address how to hold the position. The portfolio-efficiency research treats “gold” as an exposure; the structural difference between an ETF share, a futures contract, and directly-titled physical metal does not enter the math. That is a separate structural question, addressed at length on /etf-vs-physical and in Office Note 04.
The research establishes the role; the brief establishes the form.
The structural argument does not have a forecast to fail.
A structural argument is one whose validity does not depend on a particular forecast about the next quarter or the next year. The portfolio-efficiency case for gold is structural in exactly that sense — it rests on properties of the asset (low correlation to financial assets in stress, durable behavioral history, no counterparty above the bar) that do not require a market view to operate.
This matters because most public commentary on gold is the inverse: tactical, narrative-driven, tied to a specific forecast about inflation or the dollar or geopolitical risk. The narrative commentary and the structural research arrive at similar destinations sometimes, but for different reasons and with different exposures to being wrong. The narrative argument fails when its forecast does. The structural argument does not have a forecast to fail.
“The model you’re using to read the screen is the one they taught you. The screen is showing you something else. Once you take the structural argument seriously, every short-term prediction becomes background noise.”
Eric Roach · Co-founder
The research is one input to a Strategy Brief, not the conclusion.
The portfolio-efficiency research is one input to a Strategy Brief, not the conclusion. The brief takes the research as the argument that a directly-titled physical reserve has a defensible structural place in a long-horizon portfolio. It does not propose a specific allocation percentage; that decision sits with the client’s wealth advisor, who knows the rest of the balance sheet.
What the brief does is name the form, the custody, the titling, and the exit path that make the position institutional — that make the small allocation actually do the work the research says a small allocation can do.
A 4% allocation held in a pooled-allocation account at a custodian whose statement does not name specific bars is not the same instrument as a 4% allocation held directly-titled at an institutional depository under a depository storage agreement. The empirical research describes the behavior of the underlying metal; the brief describes the structural form that lets a household actually hold the underlying metal.
“The research tells us a small allocation belongs in the portfolio. The brief tells us what ‘allocation’ means in writing — what bars, at what facility, titled to whom, with what exit. Without that documentation, the position is an idea. With it, the position is an asset.”
Jose Gomez · Co-founder
Primary sources for an advisor or counsel reviewing the structural argument.
- Ibbotson, R. & Idzorek, T. (2010). “Strategic Asset Allocation and Commodities.” Ibbotson Associates / Morningstar. Establishes the efficient-frontier improvement from a small commodity allocation in a US-investor portfolio.
- Ibbotson SBBI Yearbook, current edition. The canonical long-run capital-markets dataset, 1926 to present. Tracked under Morningstar after Ibbotson’s 2006 acquisition.
- World Gold Council (2024). “Gold as a Strategic Asset, US Edition.” Updated annually. Places gold in a strategic-allocation framework with methodology notes.
- Markowitz, H. (1952). “Portfolio Selection.” Journal of Finance, 7(1). The foundational framework for thinking about correlation-weighted portfolio construction.
- Sharpe, W. (1966). “Mutual Fund Performance.” Journal of Business. Defined the Sharpe ratio, the standard measure of risk-adjusted return used throughout the WGC and Bridgewater work.
- Bridgewater Associates. All Weather framework. Methodology described in published Bridgewater papers and Ray Dalio interviews; the holding-aware structure includes physical gold among reserve-class assets.
The Office reads these alongside, not against, the public-facts panel on /signal. The portfolio-efficiency case answers why a reserve allocation belongs in the portfolio at all. The signal panel answers what the structural environment around it is doing now. The Strategy Brief is the document that puts both together for the specific household.
The next note follows the same structural reading.
Office Note 03 reads the central-bank reserve data — net buyers in aggregate every year for more than a decade — and what that institutional pattern says about how a household should think about its own reserve discipline.
The reviewed Physical Reserve Strategy Brief is delivered within five business days of intake. The brief is the document that takes the research above and applies it to your situation specifically.
Begin a Strategy Intake →