Why has gold fallen almost 30% from its record despite war and high inflation?
Alpha · deepseek-v4-pro
The short answer: gold did not fall *despite* war and inflation — it fell largely because the war and the inflation prints translated into a higher return on safe dollars and a stronger dollar, which is exactly what makes a non-yielding metal expensive to hold.
The fall is real, but it matters which measure is used. Spot gold peaked around $5,595 on January 29, hit a low near $3,986 on October 23 — about 29% below the record — and sits around $4,049 on October 28, still roughly 28% below. The London benchmark peak was lower, around $5,405, so on that measure the drop is closer to 25%. This was not one long slide; gold fell into mid-June, recovered into late August, then fell again from September.
The core mechanism is opportunity cost. Gold pays no interest or dividend, so it competes with safe assets that do. What matters is the *real* return after expected inflation. The 10-year TIPS yield rose from about 1.84% a year earlier to 2.95% by early October — its highest close since 2008. The nominal 10-year Treasury yield reached 5.31%, while the market’s expected inflation measure stayed almost unchanged at about 2.36%. That means the rise in yields was overwhelmingly a rise in *re…
Beta · gpt-6-sol
Beta: I agree that higher yields and a stronger dollar help explain September’s decline. I would not make them the main explanation for the *entire* fall from January’s peak. The World Gold Council’s first-half model assigns 24% of gold’s monthly price variability to momentum, 17% to risk and uncertainty, 14% to foreign exchange, and just 3% to rates; it leaves 30% unexplained. Those are contributions to *variability*, not percentages of the price drop. By contrast, its September analysis explicitly identifies rising Treasury yields, a stronger dollar and reduced futures positions as major contributors to that month’s 8.5% decline. (gold.org)
That distinction matters to the reader’s “almost 30%.” The WGC records a $5,405 London PM benchmark high on January 29, while describing an intraday price above $5,500; those are different measures. The manuscript’s roughly 29% figure compares a reported intraday high with the reported October 23 intraday low, not two London benchmark prices. I could not independently verify its precise October 23 low or October 28 spot quote from a primary p…
Gamma · gemini-3.1-pro
Gamma: I have verified the remaining key data points from the manuscript, and they strongly support the economic mechanism described, even if the "almost 30%" figure relies on a specific intraday-to-intraday comparison as Beta noted.
First, the manuscript's claims regarding the bond market's pricing of inflation and real yields are accurate. I checked the Federal Reserve Economic Data (FRED) series for October 5, 2026. The 10-year nominal Treasury yield (DGS10) did indeed close at 5.31%, with the 10-year TIPS real yield (DFII10) at 2.95%—its highest level since late 2008. The implied breakeven inflation rate was roughly 2.36%. This confirms the manuscript's argument: the spike in yields was almost entirely driven by real rates rising, not by long-term inflation fears. The market did indeed expect the inflation spike to pass. (richmondquant.com / FRED data)
Second, the manuscript's claim about the historical correlation between gold and real yields breaking down is also correct. RBC Wealth Management's measure puts the statistical correlation between gold and the 10-year TIPS yield at roughly 84% for the…
Researcher · gpt-6-luna
Researcher — verification notes
The strongest primary evidence retrieved supports a multi-cause, multi-episode explanation, not the claim that real yields explain the entire decline.
- The WGC’s first-half model supports the distinction between the spring and autumn. It attributes 24% of first-half price variability to momentum, 17% to risk and uncertainty, 14% to foreign exchange and 3% to rates; 30% is outside the model. These are contributions to *variability*, not shares of the total price fall. Its September commentary separately says higher Treasury yields and a stronger dollar were major contributors to that month’s decline.
Sources: https://www.gold.org/goldhub/research/gold-mid-year-outlook-2026 ; https://www.gold.org/goldhub/research/gold-market-commentary-september-2026 (gold.org)
- The near-30% figure depends on comparing intraday extremes. The WGC gives a January 29 London PM benchmark high of $5,405, while distinguishing it from an intraday price above $5,500. I did not find a reliable primary price-feed source in this search that verifies the manuscript’s exac…
The Arbiter · gpt-6-sol
The voices agree on the basic mechanism: gold pays no interest, so higher inflation-adjusted bond yields can make it less attractive. The Fed’s September 16 rate increase is confirmed by its own statement. But they differ on how much of the nearly 30% decline that mechanism explains. (federalreserve.gov)
Beta and the Researcher make an important distinction for the manuscript: the World Gold Council describes mixed interest-rate effects in the first half of 2026, when momentum also mattered. Its model assigns rates 3% of *first-half price variability*—not 3% of the total price fall. That limits Alpha’s claim that rising real yields were the main explanation throughout the year, without negating the stronger yield-and-dollar account offered for September. (gold.org)
Three checks still affect what can safely be published:
- “Almost 30%” needs a named price measure. The WGC distinguishes its $5,405 January 29 London benchmark high from an intraday…
The Arbiter's Final Verdict · gpt-6-sol
Verdict
The manuscript’s central premise stands, but its lead explanation is too broad. War and high inflation do not guarantee a higher gold price. A rise in the return on safe bonds can weigh on gold, which pays no interest; a stronger dollar can add pressure. The evidence checked here supports those forces particularly for September 2026, not as a demonstrated explanation for the entire fall from January’s record.
“Almost 30%” is a reported intraday-peak-to-intraday-low comparison, not the fall from the record to the October 28 price or a fall measured consistently with London benchmark prices. The exact spot extremes and October 28 quote were not verified against one primary feed here. The World Gold Council (WGC) mid-year outlook does confirm a different measure: its London PM benchmark high was $5,405 on January 29.
What the checks support—and what needs changing
| Manuscript claim | Finding |
|---|---|
| The first-half decline was largely an unwinding of a rapid rally. | The WGC’s first-half model gives momentu…
