Relief at the well, not at the pump
Brent settled at $99.25 (-1.09%), its first finish under $100 since 8 September
Brent broke $100 and our loading data concedes the relief is real — but diesel rose the same day, and Washington is weighing an export ban.
The crude relief is now real — our own loading data, which declined to confirm it yesterday, closed most of its gap in a single print, and we mark that against ourselves. The sharper point is where the relief stops. Brent has lost 8.7% in five sessions; diesel, the fuel that moves freight and food, rose on the day Brent broke $100, still trades roughly $108 a barrel above crude, and Washington is now weighing a ban on exporting it. The equity market has priced a disinflation off the barrel. The inflation that consumers and central banks actually meet is priced off the refined product, and that price has barely moved — which is why a Fed official can argue one hike may not be enough on the same day the Nasdaq sets a record.
Step back and the September energy shock is migrating rather than ending. It began at the chokepoint, moved into the barrel, and is now leaving the barrel for the refinery and the grid — diesel margins near historic extremes, gas up on the session, copper bid from the physical side. Each stage looks like relief to whoever was watching the previous one. That matters because the institutions that set the discount rate respond to the last stage, not the first: central banks read the consumer basket, and the consumer basket is built from products. **Falsification.** The view is wrong if the diesel margin over Brent falls below $95 a barrel within two weeks without an export ban being imposed — that would say refining capacity is catching up and the product layer is following crude down on a lag. It is also wrong if the month-end reading of the core price index for August comes in below its 3.3% prior, which would say the product squeeze is not reaching the core. Conviction: medium — the margin evidence is strong, the pass-through is inference.
The Executive Note
Brent settled at $99.25 on 2026-09-22, -1.09% on the session and -8.74% across five settled sessions from $108.75 on 2026-09-15 — its first settlement under $100 since 8 September. WTI's November contract finished $90.52 (-2.00%). The Nasdaq closed at a record 27,244 (+0.45%); the Dow fell 0.36% and the S&P 500 finished flat . Copper rose for a fifth session and natural gas settled +4.55%.
We owe the reader a mark before anything else. Yesterday this desk argued that our chokepoint departure series declined to confirm the oil decline: Hormuz sailings ran 31.6% below their thirty-day mean and the US Gulf Coast 33.3% below. We set the test in the same edition — both lanes back at their means inside two weeks with Brent under $100 would retire the view. The next settled print has Hormuz at 286 against 302.7 (-5.5%) and the Gulf Coast at 56 against 55.0 (+1.8%), with Brent at $99.25. One day is not two weeks, and daily counts are noisy in both directions. But the direction is against us, and Bloomberg reports oil traders adding bearish bets at a record pace as disruptions ease. The physical premise of the relief now looks largely correct, and we say so plainly.
That concession sharpens rather than retires the question, because the relief has not travelled. New York Harbor diesel settled $4.9421 a gallon, +1.08% on the day Brent fell — the refined barrel moving against the crude one. From its 2026-09-15 high, the highest settlement in a year, diesel is down 6.1%, against 8.7% for Brent. Priced per barrel, diesel stands about $108 above crude, against $112 on 2026-09-15. The refinery is keeping most of the gift. And Washington has noticed: the President has said he would back a ban on diesel exports as fuel prices soar . A government does not reach for an export ban when a shortage is easing; it reaches for one when the shortage has become domestic and political.
Argued rather than asserted: could diesel's resilience be a speculative squeeze that will unwind on its own? The positioning data says not. Managed-money net length in NY HARBOR ULSD fell 38% between 2026-09-01 and 2026-09-15 — the crowd was selling into the peak, not driving it. Had the product premium been a futures-market artefact, that is the observation that would have looked different. What remains is physical: refining capacity, distillate stocks, and the export pull that the proposed ban is designed to cut.
The consequence is a split in how the day's energy news is being read. The equity market has priced a disinflation off the barrel, and a crude benchmark under $100 supports it. The Federal Reserve reads the consumer basket, which is built from products, and Richmond Fed president Barkin said on the same session that one rate increase may not be enough. Traders are nonetheless buying protection against a shallower hiking cycle. The long end has chosen neither side: the thirty-year settled 5.30%, -6 basis points across the whole five-session oil decline. Europe is already living the product version — eurozone consumer sentiment was knocked by surging energy prices — and a US export ban would send more of that squeeze across the Atlantic.
Copper tells the same story from the metals side, and it comes with a second mark against us. On 15 September we wrote that Chinese production beating while investment contracted would leave a weaker floor under the metal. Production beat, 5.2% against 4.8%; year-to-date investment contracted -7.2%. Copper then rose 6.14% in five sessions to $6.7595, within 0.65% of its one-year high. Bloomberg attributes the move to tight Chinese supply, and our positioning data is consistent with a physical rather than speculative bid: managed money held 65,106 contracts net long on 2026-09-15, down 20.8% in a week and the lightest since late July. The rally started from a light crowd. Whether the crowd has joined it since is what the 25 September report will show.
**What would prove this wrong.** The diesel margin over Brent falling below $95 a barrel within two weeks without an export ban would say refining is catching up and the product layer is following crude on a lag. US August core PCE on 30 September printing below its 3.3% prior would say the product squeeze is not reaching the core. Conviction is medium: the margin evidence is observed, the pass-through to the central bank is inference.
**Method.** Tape figures are the settled session of 2026-09-22; the US cash session of 2026-09-23 had not settled when this was written and is not quoted. Brent, copper, natural gas and diesel are quoted at exchange settlement. The diesel margin converts the New York Harbor contract at 42 gallons a barrel and subtracts Brent; it is an indicator of refining tightness, not a refiner's realised margin. Positioning is read from the maximum-open-interest contract market per report date. Chokepoint figures are settled-dated and cover the lanes with live coverage only.
What mattered
Our loading data now confirms the crude relief, and yesterday's reading is marked against us
On 22 September this desk argued that departures through the two tracked oil lanes declined to confirm the oil decline: Hormuz at 210 sailings against a thirty-day mean of 307.1 (-31.6%), the US Gulf Coast at 37 against 55.5 (-33.3%). The next settled print, for 2026-09-22, reads Hormuz 286 against 302.7 (-5.5%) and the Gulf Coast 56 against 55.0 (+1.8%).
We set the test ourselves: both lanes back at their thirty-day means inside two weeks with Brent under $100 would retire the view. One lane is at its mean, the other within six per cent of it, and Brent settled at $99.25. One day is not two weeks, and daily departure counts are noisy — but the direction is against us, and Bloomberg reports oil traders adding bearish bets at a record pace as disruptions ease. The honest mark is that the physical premise of the relief now looks largely correct.
The read —What survives from yesterday is the freight observation and the long end's indifference; what does not survive is the claim that the cargo was missing. The more useful question has moved one step down the chain, to the refinery.
Diesel rose on the day Brent broke $100, and the margin between them has barely narrowed
New York Harbor diesel settled $4.9421 a gallon, +1.08% on a session in which Brent fell 1.09%. From the 2026-09-15 high of $5.2620 — the highest settlement in a year — diesel is down 6.1%, against 8.7% for Brent over the same five sessions.
Converted to a barrel, diesel stands about $108 above Brent, against $112 on 2026-09-15: the margin has narrowed only 3.5% while crude fell by more than twice that. Relief that stops at the refinery gate does not reach trucking, farming or the inflation print. The speculative crowd is not what is holding diesel up — managed-money net length in NY HARBOR ULSD fell 38% between 2026-09-01 and 2026-09-15, into the peak — which leaves physical scarcity as the likelier explanation.
The read —The export-ban discussion is the tell that this is now a domestic political problem, and Mexico's press is already describing diesel rather than crude as the country's energy problem. Name what would look different if the relief were complete: the margin would be falling faster than crude, not slower.
The Fed is reading the product, and the market is pricing the barrel
On the same day the Nasdaq closed at a record, Richmond Fed president Barkin said one rate increase may not be enough to contain inflation, while Bloomberg reports traders buying protection against a shallower hiking cycle. The long end was unmoved: the thirty-year settled 5.30%, -6 basis points across the whole five-session oil decline.
The two positions are reading different energy prices. A shallower cycle is the natural inference from crude under $100. A Fed official arguing for more is the natural inference from diesel, electricity and the consumer basket, which have not followed. We think the second reading is the one the central bank will act on, because the mandate is written against the price consumers pay, not the price traders quote.
The read —The first hard test is US August core PCE on 30 September against a 3.3% prior. A print at or above that makes the shallower-cycle protection look early; a clear miss says the product squeeze is staying in the headline and the market's reading is right.
What we see that the tape doesn't
Managed-money net length in copper (COPPER- #1, the largest open-interest market): 65,106 contracts on 2026-09-15, down 20.8% from 82,154 a week earlier and the lightest since 2026-07-28 (65,008). Copper has risen in every settled session since.
A rally that starts from light speculative length is being made by someone other than the speculators — and that is the discriminating observation for copper right now. Had the move been the futures crowd chasing an AI-and-grid story, the positioning data would have shown length building into it; instead the crowd was cutting at the lows, and copper went on to gain 6.1% in five sessions to within 0.65% of its one-year high. That is consistent with the physical attribution carried by Bloomberg, which reports signs of tight supplies in China. It also scores a call of ours against us: on 15 September we wrote that Chinese production beating while investment contracted would leave a weaker floor under the metal; production beat (5.2% against 4.8%), investment contracted (-7.2%), and the metal rose. The limitation is the date: positions are measured on 2026-09-15, the first day of the advance, so this establishes where the rally started, not who has joined it. The 25 September report covers the full move.
What to watch
- The diesel margin over Brent against a $95-a-barrel line — the one number that says whether the relief is reaching the refined barrel.
- Any formal step toward a US diesel export restriction, and how European and Latin American import prices respond in the days after.
- Chinese physical copper premiums and exchange warehouse stocks — the supply evidence behind the rally, independent of the futures price.
- The Nasdaq-Dow divergence after a session of +0.45% against -0.36%.
Risks on the radar
A US diesel export ban sends the product shortage across the Atlantic
high · highThe President has said he would back a ban on diesel exports. The scenario this issue does not itself argue is the second-round effect: the United States is a swing supplier of diesel to Europe and Latin America, and a ban that eases the domestic price would do it by exporting the shortage to importers already under strain — eurozone consumer sentiment is being knocked by energy prices and Mexico's press names diesel as the country's energy problem. Carried 20 September at medium probability and high impact; probability raised because the policy has moved from lawmakers' calls to presidential support.
The rhetoric escalates as the physical premium drains out
medium · severeThe President threatened Iran's annihilation absent a peace deal in his UN address, Washington threatened to shut down Iranian airlines as the Houthis tighten pressure on the oil route, and Libyan output was hit by a pipeline shutdown. The scenario is that escalation arrives into a market that has just removed the premium and a departure series that has just returned to normal — no cushion in either. Carried 21 September at high probability and severe impact; probability marked down to medium because the physical lanes have normalised, trend falling on the same evidence.
Private credit redemption queues become a standing feature
medium · highBloomberg reports that Apollo has capped withdrawals from a private credit fund again, with 14.7% of holders seeking to exit. A second consecutive cap turns a liquidity event into a pattern. The scenario is that semi-liquid private credit vehicles converge on gated redemptions across managers while public credit stays open — Sysco drew blowout demand for a $17 billion bond sale on the same day — which would mark the stress as structural to the vehicle rather than to credit. Carried 13 September at medium probability and medium impact; impact raised on the repeat gate.
The market prices a shallower hiking cycle the Fed does not deliver
medium · highTraders are buying protection against a shallower Fed hiking cycle while Richmond Fed president Barkin argues one increase may not be enough. The scenario is a re-pricing at the front end when the Fed's preferred core price measure for August lands at or above its 3.3% prior — the market having taken crude under $100 as the policy signal. Carried 19 September at medium probability and high impact; unchanged.
Copper tightness turns from a price story into an availability story
low · mediumCopper has risen in five straight sessions to within touching distance of its one-year high on reported tight supply in China, while Alibaba outlined AI data centres drawing 20 gigawatts by 2032. The scenario is that physical tightness shows up as delivery delays and premiums for grid and data-centre builders rather than only as a price. Probability is kept low because speculative length was light and the move is five sessions old. First carried under this key in the trailing month, so no trend is claimed.
— Antevo Executive Brief