Gold and silver did not fall because inflation risk receded. They fell because the people who owned them were levered and the dollar squeezed. Silver dropped harder than gold, copper went down with both, and the currency board tells the same story: the Swiss franc line moved 1.29% over five days, the forint 2.50%, the peso 3.17%, all in the dollar's favour. That is a funding event across the metals complex, not a verdict on prices. Treating a margin call as a disinflation signal is the standard error of the week after one.

The energy leg is the opposite kind of move: it is physical. US crude stocks excluding the SPR built by 2,969 MBBL in the week to 18 September, and Brent still rose 2.75% on the day. Price rising into a crude build means the bid is not in crude, it is downstream. Gasoline drew 1,686 MBBL, distillate drew 428, and the EIA is publishing on elevated crack spreads and what goes into diesel prices while also flagging the US on track for record crude production in 2026. Record output that does not cap the product curve is the clearest available evidence that the constraint sits in refining and in the barrels the US does not produce. That constraint does not clear on a quarter's notice.

Which leaves the 10y breakeven at 2.34% as the thing that is wrong. Set it against a 10y at 5.18%, a 30y at 5.47%, and fed funds effective at 3.88%. The front end is being cut into, the long end is refusing to follow, and the 10y-2y has steepened to 0.36. That is the term premium doing work the inflation compensation line is not. The Fed's balance sheet at $6,747,704m rose on the week rather than shrinking. Core CPI rose 0.976 index points in August. Unemployment is 4.1%. None of that composition belongs with a breakeven starting with a two.

So I am not choosing between the two readings on offer. The metals break is real and it is about dollar liquidity, the energy rally is real and it is about supply, and the breakeven is the instrument that has not marked. Markets have repeatedly priced energy squeezes as transitory because the base rate of transitory is high and because the CPI weighting dilutes the pass-through before it reaches the index.

The trade is short 10y breakevens against long refined product exposure, and gold at 4,154.96 is something to buy from the liquidation rather than sell into it. The forced sellers are not the ones with a view.

Gold and silver did not break because inflation risk receded. They broke because the discount rate moved, and the metals are the most duration-sensitive thing on the screen that nobody thinks of as duration. The 10y at 5.18% against a breakeven of 2.34% leaves a real yield doing real damage, and a non-yielding asset is the first place that shows up. Silver down 7.08% over five days and copper down 2.23% tell you the same thing from the industrial end. This is a rates event wearing a commodity costume.

The energy leg is where the obvious read goes wrong. Brent at 104.659 and WTI at 98.778, both up more than two and a half percent on the day, look like the start of an inflation impulse until you check what is behind them. US crude stocks excluding the SPR built by 2,969 MBBL in the week to September 18. Gasoline drew 1,686 and distillate drew 428. Crude is accumulating while products tighten. The EIA is telling you directly that elevated crack spreads are what is showing up at the pump, and that the US is tracking a record production year. That is a refining story and a positioning story. It is not a shortage of oil.

So the two moves are not in tension at all. Bonds repriced, metals liquidated, and refined product margins did what refined product margins do in late September. Breakevens at 2.34% are not asleep. They are correctly declining to extrapolate a crack spread.

What worries me is what the Fed does with a 10y at 5.18% when funds sit at 3.88% and the curve gives back only 0.36 of spread. Long real rates are tightening on their own, without a vote, and the committee will be looking at unemployment at 4.1% and payrolls adding 162 thousand in August and concluding that nothing is binding yet. Headline unemployment is a lagging read on a labor market that gives way at the edges first. High yield at 2.8 is not signalling distress, which is exactly the condition under which policy stays restrictive one meeting too long.

The bill for that arrives outside the US. The forint is down 2.50% against the dollar over five days, the peso 3.17%, the zloty 1.47%. Nothing happened in Budapest or Warsaw or Mexico City this week. What happened is that the long end of the US curve moved and capital did what capital does. Those countries will now be advised to defend their currencies with rates their own economies cannot carry, and the advice will be framed as discipline.

My call: the metals break sticks, the oil rally fades as the crude build works through the crack spread, and the 2.34% breakeven is the honest number on the page.

Silver fell 4.60% against gold's 3.03%, and copper went with them at 2.13%. That ratio is the whole story. A real-yield shock reprices gold; it does not take silver down half again as hard and drag copper along on a day crude is up 3.12%. What you are looking at is a margin call inside one complex, not a macro factor rotating.

Check the yields it supposedly came from. The 2y moved two basis points to 4.87%. The 10y moved seven, to 5.18%, and the 30y moved seven, to 5.47%. That is a long-end term premium adjustment. A real-yield event violent enough to knock 4.34% off gold in five sessions normally starts at the front, and the front did nothing.

The front end is also where the consensus is hiding in plain sight. Effective fed funds is 3.88%. Two-year is 4.87%. The curve has been carrying roughly a hundred basis points of tightening for a while now, 10s2s at 0.36. Nobody is long gold as a dovish-Fed expression and getting stopped out this week. That trade was repriced already. Selling metals because rates went up is buying a narrative the curve published months ago.

The position that is actually offside is short inflation. Ten-year breakeven at 2.34% with Brent at 104.659 and WTI at 98.778, crude stocks building 2,969 MBBL while gasoline drew 1,686 and distillate drew 428. Crude length accumulating, product tight, and the EIA flagging elevated crack spreads. That is the pump-price channel, and it is the one that moves expectations surveys. High yield OAS at 2.8pp rules out the growth-scare defence of low breakevens. Credit is not pricing anything. So 2.34% is a bet that energy does not pass through, held at a level set before energy moved.

Same book, other legs: MXN 3.17% weaker over five days, HUF 2.50%, PLN 1.47%, ZAR 1.02%. High-carry longs coming in as a group. Meanwhile EURJPY is down 1.09% and GBPJPY 1.08%, more than USDJPY's 0.32%. The funder firms while the carry legs and the levered metals unwind together. That is one position being reduced, not three views changing.

What I cannot reconcile: copper down 2.13% while crude is up 3.12%. Those two usually argue about the same thing. Either the energy bid is pure supply-side and copper is telling you about demand, or the metals move is mechanical enough to override the growth signal entirely. I lean mechanical, and the silver beta supports it, but that relationship breaking is the part of this tape that does not fit the rest.

Own breakevens against the 10y here. The gold holders have already paid. The people short inflation at 2.34% have not.