The Gold Price Outlook Bulls and Bears Both Get Wrong

Gold is sitting at roughly US$4,330 per ounce as of September 2026, more than double where it traded a few years ago. A rigorous academic valuation framework says the real price is nearly four times its long-run historical average, implying subdued or negative real returns over the coming decade.

At the same time, central banks just bought more than 1,000 tonnes for the third year running, supply cannot respond to price signals for nearly two decades, and a single regulatory change could trigger the largest institutional demand shock since exchange-traded funds arrived in 2004. Both of these things are true at once.

For anyone holding gold, or weighing whether to add it, that ambiguity is uncomfortable. The tired framing of “bull or bear” misses the structural complexity entirely, and understanding why the valuation warning and the demand thesis can coexist is the analytical work that matters for a portfolio decision right now.

This piece lays out the competing forces in enough detail that you can form a considered view on gold’s role in your own portfolio, rather than defaulting to price momentum or macro mood. Consider it a framework, not a verdict on where the gold price outlook lands.

The “Golden Constant” and what it says about today’s price

Start with a single number. The inflation-adjusted price of gold has reached US$4,587.21 in constant 2024 dollars, against a long-run average of US$1,168.47 measured across 796 observations. That puts the current real price at roughly 3.9 times its historical mean.

To understand why that figure should give you pause, it helps to know what it measures. The “Golden Constant” framework, associated with Professor Campbell Harvey, treats the real (inflation-adjusted) price of gold much like a price-to-earnings ratio treats a stock. It is a valuation gauge, and elevated readings have historically predicted weaker returns over a 10-year horizon.

The NBER Golden Dilemma research by Erb and Harvey established the foundational empirical case for using the real price of gold as a valuation gauge, demonstrating across long historical series that elevated real prices reliably predict below-average subsequent real returns.

The logic rests on three inputs:

• The real price divides the current quoted price by an inflation index, stripping out the effect of a depreciating currency.
• Over very long spans, gold’s real return tends toward zero, oscillating above and below that line with its volatility.
• When the real price sits high relative to history, the framework projects lower forward returns; when it sits low, it projects stronger ones.

Updated academic modelling from March 2024 puts the real price at 7.3 in 1982 dollars, another reading well above any comfortable band.

The Golden Constant is not the only quantitative lens applied to gold valuation; the M2 money supply model anchors fair value to the growth of broad money rather than inflation-adjusted price history, and in 2026 the two frameworks produce meaningfully different implied return forecasts.

Real price of US$4,587 versus a long-run average of US$1,168: roughly 3.9 times the historical mean.

What this tells you is uncomfortable but clear. If you buy at this level and nothing changes structurally, the probability-weighted expectation embedded in the framework is that gold underperforms inflation over the next decade. That fact belongs at the centre of any honest portfolio construction decision, not in a footnote.

How a decade of strong returns compounds the mean-reversion risk

Here is the part that trips up momentum-driven thinking. From its 2015 low through 2024, gold delivered a real return of roughly 6.3% per year, an exceptional stretch by any historical measure.

Intuition says a strong track record is reassuring. The framework says the opposite. A run of unusually high prior returns is treated as increasing, not reducing, the odds of subdued returns ahead, because it is the very thing that has stretched the real price so far above its mean. The strong decade is not separate from the valuation warning. It is the warning.

What the “Golden Constant” cannot account for: a reserve system in transition

A valuation model is only as reliable as the world it was calibrated on. That is the opening the structural counter-thesis walks through, and it is a genuine question rather than a rebuttal.

The argument runs like this. The global reserve system is undergoing a geopolitically driven reconfiguration, one that may justify a structurally higher equilibrium real price than post-1975 averages can capture. The historical series used to flag “overvaluation” was built during an era of unchallenged dollar dominance. If that era is fading, the benchmark itself may be mismeasured.

Evaluating how far dollar reserve currency status has actually declined, rather than how far it might decline, is critical to calibrating the severity of the valuation warning, because the structural thesis stands or falls on whether the historical benchmark is genuinely obsolete.

The fiscal backdrop gives the thesis weight. The United States carries roughly US$40 trillion in national debt, with about 20% of federal tax revenue now consumed by interest payments alone. Against that, global central banks collectively hold more gold than U.S. dollar-denominated assets, a milestone that would have seemed remote a decade ago.

There is a mechanical amplifier here too. Gold makes up a very small share of most investment portfolios, so even modest aggregate reallocations by institutions exert outsized upward pressure on price.

For the structural thesis to genuinely override the valuation signal, three conditions would need to hold:

• The dollar’s reserve share declines materially, not marginally.
• Central bank demand stays elevated rather than reverting.
• Institutional reallocation into gold accelerates from here.

Valuation models are calibrated on the world as it was, not necessarily the world as it is becoming.

Where this leaves you is deliberately unresolved. If the dollar’s structural role as the primary reserve asset is truly diminishing, the historical price series used to identify “overvaluation” is measuring against a benchmark that no longer reflects equilibrium. The valuation warning is real, but its severity may be overstated. Holding both ideas at once is the honest position the evidence currently supports.

Central bank accumulation as a structural demand floor

The strongest evidence for the structural case is not a forecast. It is a pattern that has already repeated for three years.

Official institutions added 1,045 tonnes of gold to reserves in 2024, the third consecutive year that central bank demand cleared 1,000 tonnes. Set that against the 2010-2021 annual average of 473 tonnes and the shift is stark: buying has more than doubled from its prior baseline.

The share data reinforces the scale. Central banks accounted for more than 20% of global gold demand in 2024, up from around 10% through the 2010s. Quarterly figures show the pace holding, with a record net 290 tonnes in Q1 2024 and a further 410 tonnes added across the first half of 2025.

The largest reported buyers in 2024 were the National Bank of Poland at 90 tonnes and the Central Bank of Türkiye at 75 tonnes. Reserve manager surveys, covering some 75 institutions, point to three specific risks being hedged:

The World Gold Council central bank research hub publishes the reserve manager surveys that underpin the demand figures cited here, including the motivations central banks report for their accumulation, covering inflation hedging, sanctions exposure, and the search for politically neutral assets.

• Inflation, fiscal expansion, and sovereign default risk, addressed by holding a hard store of value.
• Sanctions and abrupt changes to the international monetary system, cited by roughly one in four emerging-market central banks.
• Counterparty and seizure risk, avoided by holding a politically neutral physical asset with no credit exposure.

Notably, 32% of surveyed central banks expect to lift their gold holdings in the short term as they test alternatives to the dollar.

What three consecutive years above 1,000 tonnes tells you is that this is no longer a tactical response to a single shock. It has become the working baseline for global reserve management, and gold deserves to be assessed on that footing rather than as a fading crisis trade.

Why supply cannot rescue the market from a demand shock

If demand is doing the heavy lifting on price, the natural question is whether supply can respond and cool things down. The answer, built from the physical reality up, is no.

Gold supply is structurally inelastic. Unlike a factory that ramps output when prices rise, gold production cannot answer a price signal within any timeframe that matters to an investor. The evidence is in the numbers: newly mined gold has added less than 2% annually to total above-ground supply in each of the past 10 years, despite substantial price appreciation over that stretch.

Global mine production sits near record levels at roughly 3,600-3,800 tonnes per year, but it is plateauing. Three constraints compound to lock the supply response in place, in rough order of their weight:

1. Geological depletion, as the highest-grade deposits are progressively exhausted and new discoveries come in leaner.
2. Permitting and environmental delays, which stretch the timeline before a discovery can become a mine.
3. The development lead time itself, the sum of every prior constraint expressed in years.

That lead time is the crux. Across 127 mines entering production since 2002, the average span from discovery to first output was 15.7 years, and 15.2 years for gold specifically. For mines that came online between 2020 and 2023, it stretched to roughly 18 years.

A discovery made today, on an 18-year cycle, does not produce an ounce of new supply until the mid-2040s at the earliest.

The practical effect is that supply is removed as a near-term corrective mechanism. Four recent mine start-ups added only around 10 tonnes of combined annual capacity, trivial against global demand. The full burden of any price adjustment therefore falls on demand moderation, which means your analysis of gold should be almost entirely an analysis of demand.

Why recycling does not fill the gap

Recycled scrap does respond to higher prices, and more quickly than mining. But its scale is capped by where above-ground gold actually sits.

The majority of it rests in central bank vaults and institutional holdings that are functionally price-insensitive; those owners are accumulating, not selling into strength. Recycling can shave demand pressure at the margin, but it cannot come close to replicating the magnitude of new mine supply growth, so it offers no structural relief.

The regulatory wildcard and what it would mean for institutional demand

There is one more variable, and it is large enough that treating it as negligible is itself a positioning error.

Two rulebooks currently keep gold out of the core reserves of the largest financial institutions. Under Basel III/IV, gold held as allocated bullion carries a 0% risk weight for capital, but it is not classified as a High-Quality Liquid Asset and it attracts a punitive 85% Required Stable Funding factor under the Net Stable Funding Ratio. That treatment makes gold expensive for a commercial bank to hold. Insurers face a parallel problem, with gold often bucketed alongside defaulted junk bonds in reserve rules, a classification widely seen as inconsistent with its actual risk profile.

China’s insurance mandate is the first instance of regulatory reclassification at scale: a minimum 1% allocation to physical gold, phased in over three years. It is worth being precise here. There is no verified shift to a 5% ceiling, no parallel mandate confirmed in other major jurisdictions, and no Basel authority has formally reclassified gold as HQLA as of mid-2026.

To calibrate what a bank reclassification could do, the ETF precedent is instructive. The SPDR Gold Shares fund launched in November 2004 and grew to US$74-78 billion in assets by 2024, lifting investment demand from about 8% to 22-23% of total gold demand. A bank reserve reclassification could be structurally comparable in magnitude.

It would not be identical, though. Three conditions would likely make it play out differently:

• Substitution would be partial, not wholesale, as banks shifted only a slice of reserves.
• The shift would be heavily regulated and phased rather than market-driven.
• It could prove pro-cyclical, amplifying moves in both directions.

Market analysts assign a greater-than-zero probability to eventual reclassification, pointing to gold’s low bid-ask spreads, deep liquidity, and resilience under stress. Even if you assess the odds of full Basel reclassification within five years at only 20-30%, the sheer asymmetry of the demand shock it would unleash means it belongs in your scenario analysis, not in the tail-risk bin most retail gold commentary consigns it to.

For readers wanting to understand the regulatory mechanics in depth, our dedicated guide to gold’s Basel III HQLA classification covers the specific NSFR treatment, the LBMA’s lobbying position, and the jurisdictional variations that will determine whether reclassification happens at all.

Holding the tension: what gold’s competing forces mean for portfolio positioning

You now have both sides of the dilemma in full. The valuation framework projects subdued forward real returns from today’s entry point. The structural thesis argues for a higher equilibrium real price than history can capture, supported by proven central bank demand, inelastic supply, and a live regulatory catalyst. Neither side is wrong, which is precisely the difficulty.

Time horizon is the variable that breaks the tie. Over a 10-year window governed by mean reversion, the valuation warning tends to dominate, and the 3.9x real-price elevation is hard to argue away. Over a 5-year window, with central bank buying entrenched and a regulatory shock possible, the structural case has more room to assert itself.

There is one characteristic of gold that does not depend on resolving that debate at all. Its correlation with equities has been roughly zero since prices were allowed to float freely after Bretton Woods, and the IMF acknowledges its credit-risk-free status and long-term balance-sheet resilience while cautioning on its volatility and conditional hedging benefits.

Like an insurance policy, a hedging asset is expected to deliver lower standalone returns, because portfolio protection, not price appreciation, is its primary job.

That insurance analogy reframes the whole return question. Three positioning implications follow:

• Size any allocation for diversification, not return maximisation, which makes it defensible even under the overvaluation scenario.
• Let your time horizon decide which thesis you weight more heavily.
• Monitor the three variables that will move the balance: the central bank demand trajectory, any Basel reclassification signals, and the pace of real-price mean reversion.

For investors ready to translate the framework above into a concrete position size, our full explainer on gold portfolio allocation examines how the overvaluation scenario and the structural thesis map to specific allocation ranges within a 60/40 and an inflation-tilted portfolio.

This article is for informational purposes only and should not be considered financial advice. Investors should conduct their own research and consult with financial professionals before making investment decisions.

Past performance does not guarantee future results. Financial projections are subject to market conditions and various risk factors, and any forward-looking scenarios described here are speculative and subject to change.