Gold's $4,000 break misreads Shanghai demand

Gold under $4,000 is supposed to settle the argument. We think it does the opposite. The break in COMEX is real, and the macro trigger is real, but Shanghai is not confirming a clean collapse in underlying demand. Shanghai Futures Exchange weekly warehouse stock data showed gold stocks at 118.4 tonnes in the week to 6 November, up 1.5 tonnes on the week and the highest reading since 14 November 2025. Official Chinese reserve data available through September still showed gold holdings rising to 2,410 tonnes, the top of their one-year range and the 23rd monthly increase in a row.

The prevailing interpretation is that October weakness was mainly a de-grossing and Fed repricing event, while physical and reserve-side signals stayed supportive rather than cracking. On the evidence, that is only partly right. World Gold Council commentary said gold fell in October on asset rotation, de-grossing and Fed rate-path repricing, with COMEX managed-money net positions down US$7 billion, or 53 tonnes, and open interest falling by more than positions. At the same time, World Gold Council data showed global gold ETF holdings rose by 19 tonnes to 4,275 tonnes and flows were still positive at US$2.7 billion in October. That tells us the sell-off was led by paper and macro. It does not tell us Shanghai was passive.

COMEX broke on rates, not on a vanishing bid

The price action itself fits a macro flush. COMEX gold settled near $4,115.00 on 9 November, up $39.20, then near $4,030.80 on 10 November, down $84.20, and near $3,999.90 on 11 November, down another $30.90. From the intraday high of $4,157.10 on 9 November to the 11 November settlement, gold lost $157.20, or 3.8 per cent.

That swing lined up with inflation data that were awkward for the rate-cut story. US CPI slowed to 3.9 per cent year on year from 4.0 per cent, but monthly CPI printed 0.4 per cent against a 0.3 per cent forecast, and core CPI was 0.3 per cent against 0.2 per cent expected. In other words, the market had a clear reason to reprice the Fed path more hawkishly.

But the same evidence weakens the usual bearish leap from lower futures to weaker demand. If investors as a group were running from gold, ETF metal would not have risen by 19 tonnes in October. If this were a replay of a broad post-2011 style liquidation, World Gold Council commentary would not be pointing out that the current drawdown is not a good parallel and that recovery conditions are “much stronger”. We should read the break first as a macro-led unwind in leveraged positioning, not as proof that the metal’s user base has stepped away.

Shanghai is carrying more metal, not less

The Shanghai evidence is not the stuff of panic selling. SHFE gold warehouse stocks reached 118.4 tonnes in the week to 6 November, up 2.4 tonnes, or 2.0 per cent, from 116.0 tonnes four weeks earlier, while COMEX gold futures fell 3.4 per cent over the same span, from $4,216.30 to $4,074.90. Over 13 weeks, SHFE stocks rose 4.8 tonnes, or 4.2 per cent, while COMEX fell 7.4 per cent. Over 52 weeks, SHFE stocks climbed 28.0 tonnes, or 30.9 per cent, while COMEX was down 0.5 per cent.

Those stocks are registered warehouse warrants, not a count of all gold in China, so they are not a direct measure of total physical demand. Even so, they show metal moving into the Chinese deliverable system while Western futures weakened. That is not the signature of a market being abandoned.

Nor was the Shanghai Gold Exchange dead on Wednesday. Shanghai Gold Exchange data showed Au(T+D) open interest at 220,528 lots, volume was 27,036 kilograms, delivery was 18,796 lots, and shorts paid longs on deferral for a second straight trading day. The contract closed at 872.82 yuan per gram, down 1.28 per cent, with turnover of 23.5 billion yuan. The mini contract also saw 17,046 lots of delivery. None of that proves tightness. It does show ongoing use of the exchange for delivery and positioning even as COMEX cracked.

On reserves, the public SAFE table available through September showed China at 2,410 tonnes, up 103.3 tonnes so far in 2026 and rising month by month through the year. Reports this week then said the People’s Bank of China added 25.8 tonnes in October, extending buying to 24 months and marking the biggest increase in three years. We should separate the two: September is confirmed in the reserve table; October is reported but not yet in that table. Either way, the official sector story is accumulation, not retreat.

The bear case still has real evidence

We should not overstate Shanghai’s message. China’s premium over London was only $3.83 an ounce on 11 November, with a seven-day average of $2.06. Four weeks earlier it was $8.58, so the premium fell by $4.75, or 55.4 per cent, while COMEX declined 2.6 per cent over the same period, from $4,107.90 to $4,000.50. The premium also hit as low as minus $8.36 in the recent range. That says local urgency has cooled.

High warehouse stocks can also mean supply is comfortable, not scarce, and the SHFE data do not prove a shortage inside China. So the bearish read is not wrong to say physical support has not overwhelmed macro selling. It is wrong only when it assumes Asian demand must be weak because COMEX is weak.

What $4,000 tests now

By 21:23 UTC on Wednesday, gold had fallen below $4,000 and analysts were warning of a move towards $3,900. We think that call is missing the asymmetry. When ETF holdings are still rising, official Chinese reserves are still rising in the latest confirmed data, and Shanghai is still carrying record deliverable stocks, a futures-driven drop can reverse faster than the bearish crowd expects.

If we are right, the important question beyond this week is not whether paper can force one more downdraft. It is whether sub-$4,000 keeps attracting real buyers in China and the official sector. A deeper, persistent discount in Shanghai, falling delivery and a halt in reserve accumulation would damage this view. If those do not appear, COMEX will have told only half the gold story.