Gold Market Commentary: Anatomy of a fall

March madness

Gold fell 12% in March to US$4,608/oz, its weakest month since June 2013. Gold lost value in all major currencies, but remains up on the year.

Our monthly attribution model GRAM captured the sentiment – but not the magnitude – of the move, attributing much of the drop to momentum factors: global gold ETF outflows, a COMEX net long unwind and a price trend reversal. Lesser contributions came from US dollar strength and yields. Our forward-looking section delves into the particulars of moves in March.

Global gold ETFs shed US$12bn (84 tonnes) during the month, led almost entirely by North America with US$14bn (-87t) and Europe with US$0.1bn (-7t). Asia’s US$1.9bn (10t) inflows were a welcome positive, and highlight how dip-buying in Asia translated into much larger fund flow but lower equivalent tonnes.

COMEX managed money net long positions dropped US$2bn (19 tonnes) in March, but retain a solid long bias.

Anatomy of a fall

• Deleveraging and liquidity dynamics, not fundamentals, led the March sell‑off in gold
• Disruptions to Middle East flows are unlikely to have had a meaningful impact on the global gold price
• There are some green shoots to resuming gold’s positive trend but short-term risks including central bank mobilisation and further deleveraging remain.

Sell what you can, not what you want

Gold’s sell‑off during the first three weeks of March was sharp, counter‑intuitive, but not unprecedented.¹ It occurred against a backdrop normally supportive for gold: elevated geopolitical tensions and renewed inflation concerns. The episode is a reminder that gold is not a contractual hedge. Prices rise only when incremental buyers exceed sellers. In March, deleveraging and liquidity needs tilted that balance in favour of sellers. Some of this activity is evident in Chart 2. 

Over the three weeks to 24 March², gold appeared to overreact to a conflict-led bounce in US real yields (Chart 3). The dollar also rose but it was modest.

Our weekly GRAM model reflected this dynamic with a 12% cumulative negative residual over the period. While real yields and the dollar undoubtedly contributed to net sales, other factors were also likely at play:

First, positioning; A reported build up in retail exposure to gold risked a flush out. COMEX Non-Reportable positions³ , often associated with retail exposure, saw a cumulative 18t net drop during the first three weeks – in line with a 22t drop in Managed Money – reflecting more institutional money.⁴ A portion of gold ETF sales would also likely have been from retail hands. Global gold ETFs lost a net 80t between the beginning of March and the 24th, with the US accounting for the bulk of those.

Second, CTA-driven selling likely amplified downside momentum. Estimated and anecdotally reported Commodity Trading Advisors (CTA) were very long heading into mid-March. They reportedly unwound positions sharply when gold broke through its 50/55-day moving average on 16 March for the first time in seven months.

Third, broader cross-asset deleveraging likely spilled into gold. Elevated margin debt relative to market capitalisation probably contributed to widespread equity selling, with all but one sector in the S&P 500 (energy) posting declines. Against that backdrop, gold was not immune to liquidation pressure. Deleveraging by multi‑asset investors – including CTAs with exposure to equities, likely generated incremental selling in gold as positions were reduced to meet liquidity needs and reduce portfolio VaR. 

Fourth, bond market dynamics reinforced the pressure. US bonds were sold on a near term inflation shock with 2-year nominal yields and breakeven rate shooting higher.

Fifth, central bank intervention, and speculation about central bank sales, may also have added to downward price pressure. A decision by The Central Bank of the Republic of Türkiye (CBRT) to use approximately 50t of gold as collateral, predominantly via swaps, may have fuelled rumours of selling.⁵ There is precedence for such activity – during the 2023 earthquake and during COVID. As a major purchaser of gold since 2017 Turkey’s decision reiterates the basic rationale for why gold is indispensable as a reserve asset during market turbulence. 

That this was liquidity driven and not a change in gold strategy is backed up by data at the US Fed suggesting increased outright selling of US Treasuries by central banks to buffer higher energy price risk was occurring in tandem.⁶

Middle East flow disruption

Disruptions to market activity in parts of the Middle East are unlikely to have had a material impact on global gold prices in March.

Travel disruptions and lower tourist footfall weighed on demand for jewellery and small bars, particularly from foreign buyers (Chart 4) Local prices moved into a deeper discount to COMEX, though the adjustment was modest (Chart 5).

Trading volumes in Dubai increased during the period, but at levels insufficient to influence international prices.

High‑net‑worth investor selling was also unlikely a feature in March. They are anecdotally mobile and many hold gold outside the region, notably in Swiss vaults. Any observed outflows seem more consistent with relocation than liquidation.

While sovereign or quasi‑sovereign activity is one channel capable of influencing global prices, there is – for now – no evidence that oil exporters used gold for liquidity during the period.

Overall, while regional disruptions may have affected local pricing and activity at the margin, they do not convincingly explain the scale or speed of the March sell‑off, which was driven primarily by financial market deleveraging.

Looking ahead: fundamentals reassert but risks remain

Some early signs of stabilisation are emerging: 

• The dollar struggled to sustain gains and failed to push meaningfully beyond recent highs, reducing one source of near-term pressure
• Early April ETF flows into gold have been positive across regions
• Options markets point to elevated near‑term hedging demand, but a more constructive bias further out the curve, suggesting investors continue to view gold favourably over a medium‑term horizon (Chart 6)
• Policy tightening is likely to be rhetorical (in the US) and expectations of hikes could get unwound quickly.⁷ Any energy driven CPI impulse is likely to result in demand destruction, limiting pass through to core inflation and reinforcing the case for an eventual dovish pivot
• Anecdotal reports of wealth management, retail and physical demand appearing on price stabilisation above key technical levels.

However, risks remain. Should the conflict keep oil prices well in excess of US$100/bbl for an extended period – given that the somewhat muted response was reportedly due to buffers that no longer exist⁸ – this could risk further cross‑asset deleveraging, yield blow-outs, or gold mobilisation by the official sector.

As such, while fundamentals remain supportive, price action in the near term is likely to remain sensitive to conflict-driven liquidity needs rather than macro signals alone.

Footnotes

¹Reasons to be bullish on gold. FT, 1 April 2026.

²We use this data as the lowest point for the gold price during the month and also as coincides with the activity reflected in the third release of COMEX positioning data for the month.

³Futures and Options.

⁴Changes in Managed Money, associated with more speculative institutional activity, are on average 2-3x larger than Other Reportable flow, on an absolute basis.

⁵TCMB Başkanı Karahan'dan altın kaynaklı işlemlerle ilgili açıklamalar | Son dakika ekonomi haberleri

⁶Foreign central banks sell US Treasuries in wake of Iran war, FT, 31 March 2026

⁷WATCH: Fed Chair Powell addresses energy price spikes, challenging job market in Harvard conversation | PBS News | 30 March 2026.

⁸The oil market did not underreact. It just had buffers | Rystad Energy | 25 March 2026.