Yields fell and the dollar was unchanged, yet gold dropped more than 4% this week. If it is not because of rates, what is the cause, and is it okay to keep holding gold?
2026-W43
Rates do not explain gold this week
The 10-24 Q&A explained gold's weakness with rates and the dollar. This week's numbers do not fit that explanation. Gold futures fell to $4,007.90 (-4.14%) while the 10-year went from 5.10% to 4.99% and the dollar index moved -0.03%. The clearest day is 10-22. The table below is calculated from weekly closing prices.
| 10-22, one day | Change |
|---|---|
| 10-year yield | 5.15% → 4.95% |
| Dollar index | -0.17% |
| SPY | -1.18% |
| TLT | +1.87% |
| Gold futures | -2.16% |
| Silver futures | -4.63% |
| Gold miners GDX | -5.80% |
| Bitcoin | -3.21% |
Gold fell on a day when yields fell and the dollar weakened. Based on the rate and dollar channel, it should have risen. The only asset that rose that day was Treasuries, while non-yielding gold, silver and Bitcoin were sold alongside tech stocks. That is why the weekly correlation between GLD and SPY came out at 0.88. What remains is a flows explanation, and there are two flows hypotheses. Both are inferred from price patterns, and the sellers have not been identified.
Hypothesis 1. Selling to raise cash
When equity losses grow, investors sell what is easy to sell and has gains. Gold fits that description. The same thing happened in March 2020.
“Sharp declines in equity and other cyclical markets on 12th March were setting off circuit breakers and triggering the selling of high quality, liquid assets very likely in order to meet margin calls.” — WisdomTree (2020-03-13)
That day the S&P 500 fell about 10% and gold fell 3%. In 2008, too, gold slid with equities from $1,012 in March through Lehman's bankruptcy to $718 in November. That decline was interpreted as selling to meet margin calls and redemptions (AJ Bell, 2020-03-18). Gold then rose 170% through August 2011 (WisdomTree, 2020-03-13). A hallmark of cash-raising sales is that once the selling ends, gold recovers ahead of stocks.
There are differences too. In 2008 and 2020 the fear gauge spiked to extremes, whereas this week's VIX peak was 20.66. It was not a market-wide dump. It is more reasonable to see it as some leveraged accounts with tech losses selling gold. This part is an inference.
Hypothesis 2. Unwinding of crowded positions
The second is gold's own supply and demand.
- Gold has fallen more than 20% since its record high on January 28. In September alone it fell -8.5% to close at $4,176, and the World Gold Council (WGC) cited rising rates and dollar, and a sharp reduction in futures positions (reports citing the WGC, 2026-10).
- Meanwhile, gold ETFs took in about $10 billion in September and a record $31 billion in the third quarter, with holdings at a record 4,256 tonnes (Mining.com, citing the WGC, 2026-10).
Speculative futures money already left in September, while hedging-motivated ETF money piled up to a record. Working backward from the late-September price of $4,176 and September's decline gives about $4,560 at the end of August. That means many Q3 ETF buyers are now at a loss. A week has passed in which the hedge failed to work, at a time when more people than ever bought gold as a hedge. Selling by disappointed holders could become the next supply. This link is an inference.
Signals that separate the two
| Hypothesis | Next path if correct | Confirming signal |
|---|---|---|
| Selling to raise cash | Gold recovers quickly once equity selling stops | Gold rises again on equity down days and the joint declines in gold and Bitcoin stop |
| Position unwinding | Further decline after breaking $4,000, regardless of equities | Gold falls even on days equities rebound, and ETF holdings fall for October in the WGC's early-November release |
If next week's FOMC strongly confirms a December hike, the classic rate channel is added and gold faces a threefold headwind. If it reduces room for a hike, the rate channel turns into a tailwind for gold, and if gold fails to rise that day, more weight goes to Hypothesis 2.
What should investors do
- Do not use gold as a hedge next week. What actually offset equity losses this week was long-dated Treasuries. If you need to prepare for the FOMC, short-term Treasuries are better than gold.
- Gold holders should set a closing break of the weekly low, $3,989.10, as a checkpoint. If it breaks, the position-unwinding hypothesis gains ground, and that is when to trim some of the position.
- The condition for buying new is recovery of the hedging function. Buy in stages after seeing gold rise twice on days when equities fall more than 1%. Do not buy just because the price looks cheap.
- Avoid gold miners and silver until gold's direction is confirmed. This week GDX fell -8.56% and silver -5.76%, about twice gold's decline. If gold slips once more, the losses grow by the same proportion.