When Gold and Silver Actually Behave as Safe Havens

In September 2022, gold was trading below $1,800, and Bloomberg Intelligence senior macro commodity strategist Mike McGlone flagged both gold and silver as “generational selling opportunities.” By January 2026, gold had touched roughly $5,590.

That gap between a credentialed warning and the outcome that followed is not a story about a bad call. McGlone read the valuation signals as they stood, and the signals were internally sound. What the price action between 2022 and 2026 actually documents is how gold and silver behave now, and whether that behaviour matches what most investors believe they are buying when they reach for a defensive metal.

The distinction matters because a metal that trades like a risk asset during liquidity-driven rallies is a very different portfolio tool from one that reliably offsets equity drawdowns.

What the data from this cycle tells you is whether the metal you own is doing the job you think it is, or whether you are holding a risk asset wearing safe-haven language.

What the 2022 warning was actually built on

The 2022 thesis was not a hunch. It was a valuation reading, and understanding its evidence base is what makes the subsequent surge analytically interesting rather than simply surprising.

By late 2022, gold was trading approximately 60% above its 60-month moving average. A 60-month moving average is the average price over the prior five years, so a 60% premium meant the metal sat far above its own long-run trend. On a pure mean-reversion basis, that is the statistically defensible moment to expect a pullback.

The valuation case rested on three distinct signals:

• Gold trading roughly 60% above its 60-month moving average, an extreme upside deviation from trend.
• Gold’s volatility relative to the S&P 500 reaching a multi-year peak, resembling patterns last seen in 2007.
• Gold’s valuation against a Treasury bond index sitting at decades-long highs.

“Generational selling opportunity” was how McGlone characterised gold and silver in September 2022, with gold then at depressed levels relative to where it would later trade.

Put together, these were the fingerprints of an asset behaving like a stretched risk trade, not a quiet hedge. When something trades that far above its trend while exhibiting equity-like volatility, the historical playbook says sell into the crowd.

Silver made the picture worse on its own terms. It had drifted away from acting as a gold proxy and toward behaving like an industrial commodity, closely linked to copper. That left it doubly exposed: vulnerable to a sentiment reversal in precious metals and to any softening in industrial demand.

Here is the lesson worth carrying forward. The signals McGlone identified were not wrong in kind, only in timing. A technical overvaluation reading tells you an asset is stretched. It does not tell you when structural demand flows will stop overriding mean-reversion logic, and in this cycle they overrode it for years.

McGlone’s 2022 thesis was a textbook application of contrarian valuation signals: an asset trading far above trend, exhibiting elevated relative volatility, and priced at multi-decade extremes against bonds, exactly the pattern a mean-reversion framework treats as a sell setup regardless of the prevailing narrative.

The forces that drove gold and silver to records the thesis did not anticipate

The 2022 valuation ceiling was not respected. It was smashed, and it was smashed because a stack of demand forces compounded in a sequence that technical analysis alone could not forecast.

The foundation was central bank accumulation. Official sector buying moved from roughly 10% of global gold demand in the 2010s to over 20% by the 2024 peak, a structural shift in the supply-demand equation rather than a temporary flourish. In 2024 alone, central banks added 1,045 tonnes, the third consecutive year above 1,000 tonnes, with 333 tonnes in the fourth quarter.

The foundation was central bank accumulation, and the strategic logic behind it runs deeper than a single cycle: central bank reserve diversification away from dollar-denominated assets has accelerated since 2022, giving gold a structural demand floor that did not exist to the same degree in prior bull markets.

That sovereign buying set a floor. What lit the fuse was the 2024 shift in Federal Reserve rate cut expectations, which turned a structural demand base into a genuine price surge by pulling exchange-traded fund (ETF) and retail flows on top of it.

Silver ran on a different engine. Its record above $121 in January 2026 was powered by a mix of industrial demand, particularly solar and electronics, and heavily leveraged speculative positioning, making its bull case structurally distinct from gold’s.

The table below breaks down the four drivers and what each one implies for how you read the metals now.

Official sector demand did ease in 2025. World Gold Council methodology puts 2025 purchases at roughly 863 tonnes, a 21% year-on-year decline, though still historically elevated. The floor softened; it did not disappear.

World Gold Council demand data puts 2025 central bank purchases at roughly 863 tonnes, confirming that even as the pace moderated from 2024’s record, the structural floor sovereign buying established over the prior three years remained firmly in place.

Global gold ETF holdings reached a record 4,189 tonnes by August 2026, with assets under management of US$615 billion.

The pace culminated in gold’s all-time high near $5,590 in January 2026 and a record close of $5,414.49 on 28 January 2026. Silver peaked between $121.60 and $121.74 the following day.

The practical takeaway is that sovereign-level buying and leveraged retail speculation built a demand stack no purely technical model could have priced. For gold specifically, that means any forward price framework has to treat central bank behaviour as a baseline variable, not a rare shock.

The correlation evidence: has the safe-haven role actually broken down?

This is the analytical core of the question, and the honest answer is that the data pulls in two directions at once. Resolving that tension too quickly is exactly the error most investors make.

What the short-term correlation data shows

By early 2026, the rolling correlation between gold and the S&P 500 was around 0.55, well above the five-year average near 0.22. Correlation measures how closely two assets move together, on a scale from negative 1 to positive 1, so a reading of 0.55 means gold was tracking equities far more than its recent history would suggest.

In specific 2024-2025 windows, up to 91% of gold’s daily price movements could be statistically explained by equity moves. That is a striking figure for an asset marketed as a diversifier.

The gold-equity correlation debate is further complicated by the inflation regime sitting underneath it: periods of rising inflation consistently produce different correlation readings than disinflationary environments, which is why a single rolling figure tells an incomplete story about how the relationship will behave going forward.

The explanation sits in market structure. ETF-dominated trading and leveraged retail flows tie gold’s short-term behaviour to the same liquidity conditions that drive equities, so when investors re-leverage into risk, gold gets pulled along.

Worth flagging without over-weighting: by early 2026, gold’s equity correlation had reached levels comparable to Bitcoin’s, a structural signal that the metal’s day-to-day trading character has genuinely shifted.

What the long-run data tells a different story

Now the rebuttal. The unconditional monthly correlation between gold and the S&P 500 from 1970 to 2026 remains near zero, approximately negative 0.03. Over the full span, gold and equities simply have not moved together in any meaningful, durable way.

More importantly, during the worst 10% of equity months, gold’s correlation turns meaningfully negative, around negative 0.35. When equities suffer their sharpest declines, gold has historically pulled the other way.

Think of this as the stress-test clause in gold’s safe-haven contract. It does not activate during routine volatility or risk-on re-leveraging. It activates under conditions of severe systemic stress.

That reframes the whole debate. The useful question is not “is gold a safe haven?” but “under what conditions does gold behave like one, and how do you know when those conditions are present?”

Historically, gold’s defensive correlation re-emerges in three settings:

• Severe equity drawdowns, not ordinary pullbacks.
• Genuine systemic or liquidity crises.
• Periods of acute dollar debasement fears.

The volatility numbers underline how much movement investors are now absorbing to hold these metals. By March 2026, year-to-date volatility was up 46% for gold and 106% for silver. Regime-awareness, not category-labelling, is the operative skill.

Vulnerabilities that record prices created and why they matter for what comes next

Record prices do not just reward holders. They set in motion the very forces that crack them, and by September 2026 those forces were visibly at work. Gold had retraced to the mid-$4,000s from its $5,590 peak, and silver to the mid-$60s from $121.60-$121.74.

The three forward-looking risks are discrete and ordered:

1. Demand destruction at price thresholds. At gold between $5,100 and $5,400, the Indian jewellery market entered severe demand destruction, trading at record local discounts to international spot. Chinese demand held up at a premium over the same window, but the divergence itself tells you physical buyers are not price-agnostic. In silver, elevated prices pushed industrial users to delay projects and re-engineer applications to cut consumption.
2. Speculative positioning unwind. Silver’s move above $120 was driven more by futures and options positioning than by fundamental industrial demand. That leaves the market exposed to margin-linked selling cascades, where a single downtick can trigger a chain of forced selling.
3. Hawkish macro reversal. The most consequential bearish risk is a Fed forced to respond to persistent inflation with higher real rates and a stronger US dollar. That scenario attacks both metals at once: it dismantles the rate-cut narrative that pulled inflows in, and it strengthens the currency the metals are priced in.

Silver’s year-to-date volatility ran up 106% by March 2026, more than double gold’s, a direct measure of how much its speculative character amplifies the ride in both directions.

Here is what the retracement actually tells you. It is not simply a correction. It is evidence that the demand destruction mechanism is working as commodity markets have always worked, converting price extremes back toward equilibrium.

Any model that assumes physical buyers will absorb whatever price the futures market sets is working from a flawed assumption, and this cycle just demonstrated why. That has direct implications for how you set position sizes and stop-loss frameworks, which a generic safe-haven thesis would never surface.

Holding gold and silver with clear eyes: what regime-aware allocation actually looks like

The most common failure here is a category error: treating gold and silver as permanently defensive assets and sizing positions on that assumption. The correlation data has shown that their defensive properties are regime-conditional, which means they require active monitoring rather than a set-and-forget allocation.

Regime-aware allocation calibrates position size and conviction to whether current macro conditions match the regime in which gold’s defensive correlation historically activates. It is a shift institutional allocators have already made, moving from static safe-haven sizing toward dynamic, regime-based sizing.

Regime indicators worth monitoring

Four signals are worth tracking as a practical checklist:

Regime indicators for precious metals allocation do not exist in isolation from the broader equity market cycle; breakout signals in the gold-to-S&P 500 ratio have historically preceded the conditions in which gold’s defensive correlation activates most reliably.

• Real interest rate trajectory. Rising real rates are the primary headwind for non-income assets; falling real rates support them.
• Systemic stress signals. Gold’s conditional negative correlation of roughly negative 0.35 during the worst equity months only activates under genuine stress, so this is the signal that most directly flips gold into defensive mode.
• ETF flow direction. Flows are the leading indicator. The contrast is instructive: January 2024 marked an eighth consecutive monthly outflow of negative US$2.8 billion, roughly 51 tonnes, against the record US$615 billion AUM by August 2026. Positioning can reverse fast.
• Central bank buying pace. With 2025 purchases down 21% from 2024, a moderating floor changes the baseline you are building on.

When these indicators line up toward tighter liquidity and lower stress, gold’s defensive properties are less likely to show up. When they point toward systemic stress or monetary accommodation, the odds improve.

Miners and royalty companies: the tactical versus structural distinction

Mining equities and royalty companies offer leveraged exposure to metal prices, but they amplify both upside and downside relative to physical metal or ETF-held bullion. During the 2024-2026 cycle, their equity-market correlation risks were demonstrably higher than physical holdings.

That repositions them as a tactical overlay, not a core defensive hedge. If you are holding miners for downside protection, you are holding the wrong instrument for the job.

The question to ask before sizing any allocation is not “is gold a safe haven?” It is “is the current macro regime one in which gold’s defensive correlation is likely to activate, and what is my evidence for that?”

What four years of price action changed, and what it did not

The most honest verdict on the 2022 thesis is neither right nor wrong. McGlone was correct on the valuation signal and incomplete on the structural demand variable that overrode it. Gold was stretched; sovereign buying simply kept it stretched, and then pushed it higher.

The net finding is clearer than the debate suggests. Gold has retained its defensive properties during genuine systemic stress, a claim the long-run data supports with an unconditional correlation near negative 0.03 and a meaningfully negative negative 0.35 during the worst equity months. But its day-to-day behaviour during risk-on liquidity phases is now correlated with equities in a way earlier cycles did not show, evidenced by the roughly 0.55 rolling reading in early 2026.

The safe-haven thesis is not dead. It is conditional, and this cycle documented exactly how conditional.

Gold now sits in the mid-$4,000s and silver in the mid-$60s, well off their peaks. As central bank buying moderates from record levels and the rate cycle evolves, the regime that drove the 2024-2026 surge may not persist.

The update to make is simple: replace the static safe-haven label with a regime-conditional defensive asset whose correlation shifts with macro conditions. That is a monitoring framework, not a fixed allocation, and it changes how you size and watch your exposure.

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.