Reading time: 11 min
Gold hit $5,589 per ounce on January 28, the highest price in the metal’s history. One month into the Iran war, it was trading near $4,100. That is a decline of roughly 27 percent during an active military conflict involving the world’s most important oil chokepoint. BullionVault’s analysis of ten major wars over the past 50 years shows that gold has never performed this badly in the first four weeks of fighting. Not during the Gulf War. Not after September 11. Not when Russia invaded Ukraine. The question everyone is asking is whether gold has failed as a safe haven. The answer, based on reporting from CNBC, Euronews, Saxo Bank, and institutional flow data, is more useful than that: gold didn’t fail. It was liquidated. And the distinction between those two things is the most important lesson the commodity market has offered in years.
The Numbers: Worst First Month of Any War Since 1973
BullionVault, the world’s largest online physical gold market, published a comparison of gold’s performance during ten major conflicts stretching back to the 1973 Yom Kippur War. The average pattern across all nine pre-2026 wars was clear: gold rose roughly 4 percent in the three months before fighting began, gained another 1.1 percent in the first week, and showed a cumulative gain of 6.5 percent one month into the conflict. The 2026 Iran war has broken that pattern completely. Gold rose 23.9 percent in the month before the February 28 strikes, one of the strongest pre-war rallies in the dataset, surpassed only by the 25.6 percent surge ahead of the Soviet invasion of Afghanistan in December 1979. But once the shooting started, the script reversed. Gold dropped 13.6 percent in the first month of fighting, according to BullionVault’s data through March 27. No other war in the 50-year sample produced a negative return in the first four weeks.
The comparison is instructive. Gold jumped 2.7 percent on August 2, 1990, the day Iraq invaded Kuwait. It moved 0.0 percent on March 20, 2003, when the US invasion of Iraq began, because that war had been telegraphed for months and was already priced in. The 2026 Iran war sits somewhere between those two cases: the strikes were anticipated on prediction markets like Polymarket and Kalshi, but the scale and the Hormuz closure were not. Gold initially rose from $5,296 to $5,423 after the February 28 strikes, according to CNBC. Then it fell. And kept falling.
Why Gold Sold Off: Three Forces at Once
The decline was not a single event. It was the convergence of three distinct selling pressures that hit gold simultaneously, each reinforcing the others in a feedback loop that overwhelmed safe-haven demand.
The first was margin call liquidation. The S&P 500 had its worst session of the year in early March as the war escalated. Korean stocks lost 20 percent in two days. European equities cratered. When equity portfolios fall that fast, investors who hold leveraged positions face margin calls from their brokers: demands for additional cash to cover the falling value of their collateral. Gold, because it is the most liquid non-cash asset in the world, becomes the first thing sold to meet those demands. Investors weren’t selling gold because they had lost faith in it. They were selling gold because they needed cash, and gold was the only thing in their portfolio they could sell instantly at a fair price at 3 AM on a Monday morning. This is the paradox at the heart of gold’s March decline: the same liquidity that makes gold valuable as a safe haven is what makes it vulnerable to forced liquidation during broad market panics.
The second force was the US dollar and Treasury yields. The dollar index (DXY) strengthened above 100 on safe-haven flows, according to multiple reporting sources. The 10-year Treasury yield climbed to 4.384 percent, according to Ainvest reporting, the highest level since before the Global Financial Crisis of 2007 when adjusted for the timing of the move. Gold pays no interest. When you can earn 4.4 percent risk-free on a Treasury bond, the opportunity cost of holding gold rises sharply. Ole Hansen, head of commodity strategy at Saxo Bank, told CNBC that gold’s failure to rally during the conflict “highlights the dominance of real yields and liquidity-driven selling over traditional safe-haven flows.” Euronews put it more bluntly: “energy inflation is outpacing safe-haven demand.” Four central banks froze rates in the same week in March, and the CME FedWatch tool showed greater than 50 percent probability of a rate hike for the first time since the tightening cycle ended. Higher rates for longer is the worst possible macro environment for gold.
The third force was profit-taking and ETF liquidation after a parabolic rally. Gold rose over 60 percent in 2025, according to CBS News and World Gold Council data. Annual inflows into gold ETFs surged to $89 billion in 2025, the largest on record, with global ETF holdings reaching a historic peak of 4,025 tonnes, according to the World Gold Council. That is an enormous amount of capital that entered gold at prices between $3,000 and $5,500. When the price began falling, the unwinding was fast and violent. The SPDR Gold Trust (GLD), the world’s largest gold ETF, recorded a single-day outflow of $2.91 billion on March 4, the largest daily withdrawal since 2016, according to FinancialContent reporting citing State Street data. GLD lost 25 tonnes of bullion in seven days, the sharpest weekly decline since July 2022. In total, gold ETFs shed an estimated $9 billion in cumulative net outflows over three weeks in March, according to MEXC research citing TheCCPress data. The iShares Gold Trust (IAU) saw outflows of $3.77 billion in the same period.
The Paradox: Gold as ATM, Not Casualty
The framing in most coverage has been “gold fails as safe haven.” The Financial Times ran the headline. Newsweek ran it. CBS ran it. But the framing misses the mechanism. Gold didn’t fail as a safe haven in the way that, say, a credit default swap fails when the issuer goes bankrupt. Gold remained liquid, deliverable, fungible, and universally accepted as collateral throughout the entire crisis. It didn’t gap down. It didn’t freeze. It didn’t close for trading. Every exchange that lists gold continued to operate normally.
What happened is that gold functioned as a source of emergency liquidity during a multi-asset margin event. Investors sold gold to cover losses elsewhere because gold was the one thing they could sell. This is not a failure of the safe-haven thesis. It is a confirmation of it, expressed through a channel that most retail investors don’t understand. The asset you sell in a crisis is the asset you trust the most, because it is the asset that will give you a fair price when nothing else will. Brent crude touched $120 and crashed to $86 in a single session on a false ceasefire rumour. Bitcoin fell toward $65,000 in the first week of the war. Korean equities lost a fifth of their value in 48 hours. Gold’s decline was orderly by comparison. The drawdown was severe, but it was continuous, liquid, and tradeable at every point. That is what a safe haven looks like when the entire financial system is under stress.
The Paper-Physical Divergence
The most telling detail in the March sell-off was the divergence between paper gold (futures and ETFs) and physical gold (bars, coins, and central bank purchases). While GLD was bleeding $9 billion in three weeks, physical gold demand in Asia remained strong. Gold ETFs in India saw inflows equivalent to $560 million in February alone, with net growth in 2026 reaching almost $20 billion, according to BullionVault citing MoneyControl data. Central banks in China, India, Turkey, and the Gulf states continued buying physical gold throughout the sell-off, maintaining the structural accumulation pattern that has been in place since 2022.
This paper-physical split is not new but the magnitude is historic. The FinancialContent analysis of the GLD liquidation compared it to the 2013 and 2016 “gold pukes” that signalled the end of multi-year bull runs. But the analysts noted a critical difference: unlike 2013, the 2026 sell-off is occurring in an environment where central bank buying provides a structural floor. The report described a tug-of-war between private institutions exiting through ETFs and sovereign buyers in Asia and the Middle East continuing to accumulate physical reserves. The sell-off is a futures market event, not a physical demand collapse. JPMorgan’s 2026 year-end gold target remains $6,300 per ounce. Deutsche Bank maintains $6,000. Both targets were set before the Iran escalation and, if anything, the fundamental case has strengthened: the war increases the incentive for non-Western central banks to hold gold as a reserve asset independent of the dollar system.
The Gold-to-Bitcoin Rotation
On March 4, the same day GLD recorded its $2.91 billion outflow, spot Bitcoin ETFs including BlackRock’s IBIT reported net inflows of over $460 million, according to FinancialContent data. Over the full month of March, Bitcoin ETFs attracted $2.5 billion in total inflows, the strongest monthly inflow since mid-2025, according to MEXC research. This happened while gold ETFs lost $9 billion. The rotation was visible in prices: Bitcoin fell initially when the war began but recovered faster than gold, trading between $71,000 and $73,500 by mid-March while gold was still falling.
This is not a story about Bitcoin replacing gold. The two assets serve different functions and attract different investor bases. But it is a story about generational and institutional shifts in how crisis capital is deployed. Younger institutional desks and crypto-native allocators used the gold sell-off as an opportunity to rotate into Bitcoin, treating it as the preferred volatility play. Brent’s record monthly gain drove capital into energy. Bitcoin’s recovery attracted crypto flows. Gold, sitting at the intersection of all three forces (margin liquidation, dollar strength, and rotation to alternatives), absorbed the selling pressure from everywhere simultaneously.
What the Banks Still See
The bearish price action in March has not changed the institutional consensus on gold’s direction. JPMorgan’s $6,300 year-end target and Deutsche Bank’s $6,000 target were both cited by CNBC and GoldSilver.com as unchanged despite the sell-off. The reasoning is straightforward: the structural drivers of gold’s 2024 and 2025 bull run, central bank accumulation, US fiscal deficits, de-dollarisation among BRICS nations, and geopolitical instability, have not reversed. If anything, the Iran war has intensified them.
The correction has cleared out what the market calls “weak hands”: leveraged positions, momentum traders, and late-cycle retail buyers who entered gold above $5,000 on the assumption that prices only go up during wars. GLD’s $9 billion outflow represents the flushing of those positions. What remains is the structural base: central bank reserves, long-term physical holders, and gold IRA allocations that don’t trade on a daily basis. The State Street SPDR Gold Strategy Team’s March 2026 monitor noted that gold’s 30-day realised volatility was running at roughly 13.6 percent, compared to approximately 25.1 percent for silver and roughly 52 percent for Bitcoin. Gold is still, by a large margin, the lowest-volatility alternative asset. It just had a bad month.
For positioning, the question is whether the $4,100 low holds or whether a second leg down takes gold below $4,000. CBS News reported on March 27 that gold at $4,433 sat more than 20 percent below its record, creating what the outlet described as an attractive entry point for prospective buyers. GoldSilver.com’s analysis identified the $5,000 level as the key threshold, arguing that as long as gold holds above it, the current decline qualifies as a correction within a larger bull market rather than a trend reversal. Gold currently trades near $4,500, which is above the March low but well below the January high. If Brent stays above $100 and Treasury yields remain elevated, the opportunity cost headwind persists. If the war de-escalates, oil falls, and the Fed signals cuts, gold rallies. The metal’s direction from here depends entirely on which macro force dominates: the inflation that argues for higher yields (bad for gold) or the recession risk that argues for rate cuts (good for gold).
What March Teaches About Gold in Crises
Every generation of investors learns the same lesson about gold during its first real crisis: gold does not go up in a straight line when wars start. It goes up before wars start, sells off when the shooting begins as leveraged positions unwind, and then recovers over the following months as the structural case reasserts itself. The 1990 Gulf War, the 2001 Afghanistan invasion, and the 2022 Ukraine war all followed this pattern. The 2026 Iran war is following it too, with the notable difference that gold entered the war more overextended than in any previous case (up 60 percent in 2025, with record ETF holdings) and therefore had more speculative positioning to flush.
The BullionVault data on wartime gold performance across 50 years makes one thing clear: the first month is not the trade. The average return one month into a conflict was 6.5 percent, but the dispersion around that average was wide. What mattered for long-term returns was not the initial shock but the duration and economic consequences of the conflict. The 1973 oil embargo, which is the closest historical analogue to the 2026 Hormuz closure, saw gold nearly double in the year following the embargo’s start, based on London fixing data. If the Iran war closes the strait for months rather than weeks, and if the resulting energy shock pushes the US economy into stagflation, gold’s March decline will look like a buying opportunity in hindsight. If the war ends quickly and oil returns to $70, the decline will look like the beginning of a deeper correction. Both outcomes are possible. Neither has been priced with any confidence.
