Crude rallying into a 95.0 MBBL build is not a contradiction, and reading it as product-side noise misses what the barrel is telling you. Distillate drew 2,228.0 MBBL and gasoline drew 2,536.0 MBBL in the same week that crude piled up. That is not a demand surprise. It is a conversion constraint. There is plenty of oil and not enough capacity to turn it into the fuels that actually move freight.

The obvious read is that this resolves itself: refiners come back from turnaround, the cracks compress, WTI at 86.941 gives back its 3.34%. I do not think it resolves. A decade of capital discipline was applied to upstream and downstream alike, and downstream got the worse end because refining carries transition risk that no board wants to underwrite for a twenty-year payback. The EIA's own note on Nigerian shipments tells you where net new capacity is going, and it is not into the Atlantic basin's traditional supply chain. Longer Permian laterals and record US gas production in 2026 solve for molecules, not for diesel. The level shift is in the crack, not in the crude price, which is why the crude price looks confusing.

What the rates market did with this is more interesting than what it did with oil. The two-year jumped 0.140pp to 4.34 against an effective rate of 3.63, and the ten-year breakeven went the other way, down 0.020 to 2.31. That combination is the front end pulling cuts out of the curve without any inflation expectation attached. The 10s2s spread flattened 0.080 to 0.39. Gold fell 4.34% over five days and silver 3.50%. Nobody is buying this as debasement, and on the evidence available they are right not to.

The Fed is currently doing the thing I usually argue it will not do. The balance sheet ran off another 14,787 million dollars in the week to August 26, the effective rate has not moved, and a ten-year at 4.73 against a 2.31 breakeven is a real yield that bites. High yield OAS at 2.6 says nothing has broken yet. Holding here is a defensible reaction function.

The test comes if distillate keeps drawing into the fourth quarter. Fuel inflation transmits through freight into core with a lag long enough that a committee can cut before it shows up and claim the data supported it. That is the sequence to watch, and the thirty-year at 5.22 suggests the long end has already priced who pays if it happens.

My position: fade the assumption that the crude build caps this. Own the products, not the barrel.

Oil at 86.941 does not warrant a single basis point of Fed response, and the 14bp the two-year added into Friday was bought with the wrong story.

Look at what actually moved. Crude stocks excluding the SPR built 95.0 MBBL while gasoline drew 2,536.0 MBBL and distillate drew 2,228.0 MBBL. That is a refining spread, not a scarcity. Barrels are sitting in tanks; the constraint is the capacity to turn them into something a truck can burn. Late August distillate draws ahead of turnaround season are the most predictable thing on the energy calendar. Meanwhile EIA is publishing that the Permian is producing more per well, that eight liquids pipelines have come online since the start of 2025, that Dangote is pushing product volumes out of Nigeria, and that the US is on track for record natural gas production this year. Natgas up 4.43% on the week in the middle of a record supply year tells you this is a positioning move dressed as a fundamentals move.

The rest of the market agrees with me, which is the part the rates desks are skipping. Ten-year breakevens fell to 2.31%. Gold is down 4.34% over five sessions. High yield OAS sits at 2.6 and tightened. Nothing in that set is consistent with an inflation scare. Breakevens down while the two-year jumps to 4.34% is a pure expected-real-rate move: the market is not pricing more inflation, it is pricing a Fed that leans against a relative price change. The curve did what it does when that happens, with 10y-2y compressing to 0.39.

The number that should worry people is not on the energy screen. Core CPI rose 0.724 index points in July against 0.245 for headline. Core running hotter than a print that includes food and fuel is not an oil story and it will not resolve itself when cracks normalise. That is where the case for holding at 3.63% actually lives, and nobody is making it because oil is the more available excuse.

That distinction matters because of who pays. Payrolls fell 23,000 while unemployment dropped to 4.1%. A falling jobless rate alongside job losses is people leaving the count, not people finding work. Tighten into that on the strength of a refining spread and the adjustment comes out of the low end of the wage distribution, which contributed nothing to the core stickiness that is the real problem. And the broad dollar at 118.7479 has already done its work abroad: the zloty gave up 0.95% and the forint 0.98% over five days on a decision taken in Washington.

My call: WTI does not hold 86.941 into the end of the quarter, the crack unwinds, and the two-year keeps the level for reasons that have nothing to do with crude.

Crude did not repice the Fed. The front end did that on its own, and it did it while breakevens went the other way.

The 2y went to 4.34% on a 0.140pp jump, the 10y to 4.73% on 0.060, 10y-2y compressed to 0.39. That is a bear flattener. Ten-year breakevens fell 0.020 to 2.31%. So the market took the front end higher and simultaneously marked down expected inflation. That is a real-rate move, a terminal-rate move, and it is not a story about a barrel of oil. If the desk believed a 3.34% day in WTI fed through to CPI, breakevens would have gone up. They went down.

The lead read has the causality backwards. Energy is not complicating the Fed. The Fed complication is already priced, in real rates, and energy is a separate ticket.

On the barrel itself: a 95.0 MBBL build in crude excluding SPR against draws of 2,536.0 in gasoline and 2,228.0 in distillate is a refining margin story. Flat price rallied because the products are tight and the crude is not. That is a spread, and spreads mean-revert into turnaround season. Meanwhile EIA has the US on track for record natural gas production in 2026 and natgas is up 4.43% on the week anyway, which tells you the bid is demand-led, not scarcity-led. WTI outran Brent on the day. Domestic pull, not geopolitical premium. There is no risk premium in this move that I can locate.

Where somebody is actually hurting: gold, down 4.34% over five days, silver down 3.50%, both bleeding into a week when the 2y rose and the broad dollar index printed 118.7479, up 0.390. Long metals against a cutting Fed was the consensus carry. The Fed has held the effective rate at 3.63% and the front end just repriced hawkish. That exit is narrow. Five percent in a week on gold is not a correction in a crowded position, it is the first leg.

CEE took the same hit through the bund. USD/PLN +0.95% and USD/HUF +0.98% on five days with DE10YB down 0.52%. Local carry unwinding on duration, not on anything domestic.

What does not fit: high yield OAS at 2.6, tighter by 0.030, with the 30y at 5.22 and the Fed balance sheet down 14,787 on the week. Credit is not pricing anything the rates market is pricing. Either spreads are late or the front end is wrong, and spreads have been the slower of the two for most of this cycle.

USD/JPY at 159.74 did nothing, EUR/JPY at 185.574 did nothing. The BoJ published a purchase schedule and payment statistics. The carry is not the event this week. Real rates are.