Real-Asset Split: Monetary Versus Physical Demand
Tracks whether the monetary real assets, gold, silver, miners, have decoupled from the physically consumed ones, energy, industrial metals, food, so that a single "real assets" basket no longer moves as one and each half answers to a different driver.
September 16 update: the split held a fifth window and narrowed to a single leg. The monetary half failed once more, the gold fund down 2.38% and miners 5.99% against a 10-year up 19bp in five sessions to 4.97%, while the consumed half kept separating internally: WTI rose 13.73% in five sessions to $104.04 with Brent $109.12 before settling $107.19, and copper fell 4.53% to $6.50, bigger than all but 24 of the last 120 such moves, a week after one-year-high closes. So energy is now the only part of the basket rising, which is a narrower fact than the monetary-versus-consumed divergence this row was opened on and points at the rates repricing marking down everything that is not a supply shock. Next check: the gold fund holding above $385 through September 25, and whether copper's give-back extends while crude holds; the basket recombining on one dollar or real-yield impulse invalidates. September 12 update: the split held a fourth window and the consumed half stopped moving as one. Gold fell into a confirmed Gulf disruption while WTI rose 8.44% to $100.05 and Brent 8.78% to $104.30, but natural gas fell 3.41% and copper gave back 1.78% to $6.47, and the co-movement measures say this is structural rather than a single tape: the 20-session correlation between natural gas and WTI ran -0.38 against a 60-session +0.24 through September 10, with Brent and gas at -0.37 on the same basis. So the monetary half stays pinned away from the complex while the consumed half now separates along a supply map of its own, European gas near EUR 74 per megawatt hour against US gas down 7.30%. Next check: measured cross-correlation between precious and consumed commodities falling further, or consumed strength reaching producer prices, freight, or importer trade balances; the basket recombining on one dollar or real-yield impulse invalidates. September 9 update: the persistence condition is effectively met. Copper closed $6.85 and $6.81 on September 8 and 9, the highest in a year, with London warehouse stocks down for 42 straight days through mid-August, while Brent went through $100 and WTI held $91.48, and against that the monetary half fell, GLD -0.24% into a 0.80% dollar decline with length cut 20,024 to 222,188 (report date September 1). Three consecutive windows, spanning both signs of the dollar, is the stated three-week separation on daily evidence rather than one shock week. The demand mix inside copper points the same way: electric-car sales grew about 2% and solar installations are forecast to fall, so grid construction rather than consumer goods is pulling the metal, which is a physical-demand channel the monetary half has no access to. Next check: measured cross-correlation between precious and consumed commodities falling rather than the divergence resting on weekly percentage moves; the basket recombining on a single dollar or real-yield impulse invalidates. September 5 update: the divergence survived a reversal in the driver that could have explained it away. WTI rose 8.02% and natural gas 3.48% while GLD fell 0.24%, SLV 0.94%, and GDX 0.02%, and unlike last week the dollar fell 0.54%, so the consumed half rallied hard and the monetary half did not respond to an impulse that should have helped it, with the 10-year real yield unchanged. Two consecutive weeks is one short of the three-week persistence trigger, and the composition changed between them, which is the stronger version of the observation. Next check: a third consecutive week of separation, or measured cross-correlation between precious and consumed commodities falling rather than the divergence resting on weekly percentage moves; the basket recombining on a single dollar or real-yield impulse invalidates. August 29 update: the split showed up inside a single week. Gold fell 4.59% and silver 4.44% purely against a 0.85% dollar rally, while natural gas rose 5.71%, copper held roughly flat at -0.47%, and strength was also reported in the food complex. No standing row holds this: the metals rows score gold against the dollar and real yields, the weather row scores crops against El Niño and European drought, and the industrial-demand question was closed when copper reclaimed its January high, but none of them carries the observation that the two halves have stopped co-moving. Carry it as its own row until the basket either recombines or the divergence persists.
What would move it
▲ Raise
- The divergence persists for three or more consecutive weeks rather than resolving inside one dollar move.
- Energy, industrial metals, and softs move together on physical supply news while precious metals stay pinned to the dollar and real yields.
- Cross-correlation between precious and consumed commodities falls measurably rather than anecdotally.
▼ Lower
- The basket recombines and both halves move on the same dollar or real-yield impulse.
- The week reads as a one-off dollar shock, with gas and food strength traced to weather or storage rather than to a separate demand channel.
Live triggers
- Next check: gold ETF holds above $385 through September 25.partial
Conviction over time
Event timeline
- measured October 12, 2026
Dollar Index (DXY) 102.58 (2026-10-12): the highest close in a year.
- measured October 7, 2026
Dollar Index (DXY) 102.24 (2026-10-07): the highest close in a year.
- measured October 5, 2026
Dollar Index (DXY) 102.17 (2026-10-05): the highest close in a year.
- measured October 5, 2026
10Y Yield 5.31% (2026-10-05): the highest close in a year.
- measured October 1, 2026
Dollar Index (DXY) 102.10 (2026-10-01): the highest close in a year.
- measured September 30, 2026
10Y Yield 5.29% (2026-09-30): the highest close in a year.
- measured September 29, 2026
Real-yield squeeze on gold: Treasury yields rose while expected inflation did not, so the inflation-adjusted return on bonds went up and gold, which pays no interest, paid for it. (10Y +16bp in five sessions, 10Y BEI +0bp in five sessions, GLD down about 1.4 times its usual five-session move)
- measured September 29, 2026
Dollar up, metals down: The dollar rose sharply over five sessions while gold fell: gold is trading off the dollar and interest rates, not its own story. (DXY +0.77% in five sessions, GLD -4.66% in five sessions)
- measured September 28, 2026
Dollar up, metals down: The dollar rose sharply over five sessions while gold fell: gold is trading off the dollar and interest rates, not its own story. (DXY +0.75% in five sessions, GLD -4.54% in five sessions)
- measured September 28, 2026
10Y Yield 5.18% (2026-09-24): the highest close in a year.
- measured September 27, 2026
10Y Yield 5.18% (2026-09-24): the highest close in a year.
- measured September 26, 2026
10Y Yield 5.18% (2026-09-24): the highest close in a year.
- measured September 17, 2026
10Y Yield 5.00% (2026-09-15): the highest close in a year.
- measured September 16, 2026
10Y Yield 4.97% (2026-09-14): the highest close in a year.
- lower September 16, 2026
Evidence: 10Y Yield 4.97%: up 19bp (0.19 percentage points) in five sessions, bigger than all but 5 of the last 120 five-session moves, and the gold miners fund fell 5.99% on the week even while oil surged.
- measured September 15, 2026
10Y Yield 4.96% (2026-09-11): the highest close in a year.
- lower September 15, 2026
Evidence: 10Y Yield 4.96%: up 19bp (0.19 percentage points) in five sessions, bigger than all but 3 of the last 120 five-session moves, and over the same week the gold ETF fell 2.38% instead of attracting a safety bid.
- measured September 14, 2026
10Y Yield 4.95% (2026-09-10): the highest close in a year.
- measured September 12, 2026
10Y Yield 4.95% (2026-09-10): the highest close in a year.
- measured September 11, 2026
10Y Yield 4.83% (2026-09-09): the highest close in a year.
- measured September 10, 2026
10Y Yield 4.80% (2026-09-08): the highest close in a year.
- measured September 9, 2026
Copper $6.81 (2026-09-09): the highest close in a year.
- measured September 8, 2026
Copper $6.85 (2026-09-08): the highest close in a year.
- lower September 7, 2026
Evidence: From stored daily closes, USD/JPY 154.46 (2026-09-07): the lowest close in about six months (141 sessions) , and the dollar index fell 0.80% on the week, yet the gold fund still slipped 0.24% and gold bets were trimmed by 20,024 contracts (official latest, CFTC report date September 1, 2026).
Related claims
| Opened | Claim | Status |
|---|---|---|
| October 14, 2026 | U.S. CPI 2026-10-14 printed hot versus the Cleveland Fed nowcast; the map said Gold (GLD) down on the day (settled on the release-day move in stored daily closes). | correct |
| October 14, 2026 | U.S. CPI 2026-10-14 printed hot versus the Cleveland Fed nowcast; the map said 10Y Yield up on the day (settled on the release-day move in stored daily closes). | awaiting outcome |
| October 14, 2026 | U.S. CPI 2026-10-14 printed hot versus the Cleveland Fed nowcast; the map said Dollar Index (DXY) up on the day (settled on the release-day move in stored daily closes). | correct |
| October 6, 2026 | 3-Year Note Auction 2026-10-06 printed hot versus the prior same-term auctions; the map said 10Y Yield up on the day (settled on the release-day move in stored daily closes). | incorrect |
| October 2, 2026 | Employment Situation 2026-10-02 printed cool versus a fixed threshold; the map said 10Y Yield down on the day (settled on the release-day move in stored daily closes). | incorrect |
| September 28, 2026 | Retail Sales 2026-09-28 printed hot versus a fixed threshold; the map said 10Y Yield up on the day (settled on the release-day move in stored daily closes). | correct |
| September 28, 2026 | Retail Sales 2026-09-28 printed hot versus a fixed threshold; the map said Dollar Index (DXY) up on the day (settled on the release-day move in stored daily closes). | correct |
| September 16, 2026 | Gold (GLD) closes above 390.00 on 2026-09-23 (settled by stored daily closes). | correct |
| September 16, 2026 | Retail Sales 2026-09-16 printed hot versus a fixed threshold; the map said Dollar Index (DXY) up on the day (settled on the release-day move in stored daily closes). | correct |
| September 16, 2026 | Retail Sales 2026-09-16 printed hot versus a fixed threshold; the map said 10Y Yield up on the day (settled on the release-day move in stored daily closes). | excluded |
| September 15, 2026 | Gold (GLD) closes above 385.00 on 2026-09-25 (settled by stored daily closes). | correct |
| September 14, 2026 | U.S. PPI 2026-09-14 printed hot versus a fixed threshold; the map said Dollar Index (DXY) up on the day (settled on the release-day move in stored daily closes). | correct |
| September 10, 2026 | U.S. PPI 2026-09-10 printed hot versus a fixed threshold; the map said Dollar Index (DXY) up on the day (settled on the release-day move in stored daily closes). | correct |
| September 7, 2026 | Gold (GLD) closes above 400.00 on 2026-09-11 (settled by stored daily closes). | incorrect |
| August 12, 2026 | U.S. CPI 2026-08-12 printed cool; the map said Dollar Index (DXY) down on the day (settled on the release-day move in stored daily closes). | incorrect |
| August 12, 2026 | U.S. CPI 2026-08-12 printed cool; the map said Gold (GLD) up on the day (settled on the release-day move in stored daily closes). | correct |