The number on your screen isn't "the price of oil." It's the price of oil somewhere specific, net of everything standing between the barrel and its buyer: ships, tanks, pipes, insurance paperwork, and permission. This primer distills how that middle layer actually works, and why it's where the money lands when the world breaks.
Spring 2020: demand had cratered by ~20m barrels a day while the Saudis and Russians flooded the market. But cheap oil doesn't explain a negative price. What does is a tank farm in Cushing, Oklahoma.
A WTI futures contract is just a number on a screen — until expiry day, when whoever holds it has promised to take physical delivery of 1,000 barrels at Cushing. By 17 April, roughly 60m of Cushing's 76m barrels of working storage were full, and every remaining gallon was leased or spoken for. A crowd of paper longs woke up holding delivery obligations in a town with no room.
Brent settled positive that same afternoon. Deferred WTI contracts never went negative. Nothing was wrong with oil. The middle was full.
Rerouting is a demand shock with no new demand: the Baltic-to-Rotterdam run is ~1,250 nautical miles; the same cargo to India's west coast is ~8,750. Distance beats volume. In shipping, demand is barrels × distance ("tonne-miles"), and sanctions manufactured one of history's biggest tanker employment schemes out of pure geometry.
Every barrel sits on a ladder of deliverable prices. It's worth one price if it can reach a benchmark-quality buyer, less if it can only go regional, less still if it's stuck behind a full pipeline, and nothing if it can't legally or physically move. Barrels always seek the cheapest machine with spare capacity — and when that machine fills, the price gap widens until the next, more expensive machine turns on.
One law, and it explains Alberta's $40 discounts, Waha's negative gas, and every regional blowout in modern history. Tap each rung:
Midstream assets earn in two modes. A toll is a contracted pipeline or take-or-pay terminal: volume × fee, indifferent to oil price, valued like infrastructure debt. An option is a spot tanker, an uncommitted tank, a flexible book: worth little in calm markets and worth anything in broken ones. Confusing the two is an expensive mistake — tanker equities are effectively listed call options, which is why they look cheap at every top and broken at every bottom. Buying them off a P/E screen is the classic trap.
Everything in the middle gets paid for closing one of three gaps: where a barrel is vs. where it's wanted, when it exists vs. when it's needed, and what it is vs. what the buyer's equipment can digest. A VLCC, a salt cavern, and a diluent contract are the same business pointed at different gaps. (A fourth gap — barrels that may legally move vs. those that may not — is the sanctions trade.)
Freight & tariffs. Tanker freight is quoted in Worldscale points — a taxi meter recalibrated once a year. WS100 is the published flat rate for a route; WS310 (where Gulf-to-China fixed in late June 2026) means the meter is running at 3.1× its calibrated norm. What the owner actually earns is the TCE: gross freight minus bunkers, port and canal fees, divided by round-trip days — including the empty ballast leg home.
A route is open when the price gap between two locations exceeds the cost of covering the distance. That single sentence is the entire theory of oil arbitrage.
Storage & the curve. The futures curve is the published price of time. In contango (deferred above spot) the market pays anyone who can warehouse the surplus: buy spot, sell the deferred future, store, deliver. In backwardation (spot above deferred) storage earns nothing and inventories drain.
Spring 2020 blew the spreads out past tank rent, past financing, until they covered the most expensive warehouse on earth: VLCCs hired to sit still. Floating storage peaked near 200m barrels. Euronav booked $485m of net profit in H1 2020; Vitol cleared $3bn for the year — the same event that printed −$37.63, seen from the winning side.
Blending & the spec cliff. A crude grade or refined product is a pass-fail exam. Marine fuel at 0.50% sulphur is a compliant commodity; at 0.51% it's a different, cheaper product with a smaller buyer pool. Price doesn't slope across that boundary — it steps.
Canada's bitumen is too viscous to pipe at all, so every barrel of Western Canadian Select embeds a form trade: ~30% purchased condensate. And IMO 2020 — one UN regulation about smokestacks — repriced a refining stream, a blending complex, and the relative value of every large ship on earth overnight.
The curve's shape is the market's own signed daily statement about whether the physical present is heavy or tight. Toggle between the two regimes:
"Brent" is assembled daily by named organizations out of a small set of physically deliverable cargoes under written eligibility rules. The North Sea stream now loads so little that Texas cargoes delivered into Rotterdam are netted back to a notional North Sea price using a published freight adjustment — the world's oil reference literally contains a tanker assessment as a term in its equation.
When war shut Hormuz in March 2026, Platts suspended any Dubai-deliverable grade requiring strait transit. The pool collapsed from five grades to two — Oman and Murban, the two that load outside the strait. Benchmark eligibility followed the bypass map. Dubai printed a record $157.66, and some Asian refiners re-papered purchases off ICE Brent because their benchmark had become unusable.
Sort them by substitutability, not size. Malacca is the biggest by volume but has workarounds. Hormuz has ~4.7m bbl/d of pipeline bypass against a 20.9m flow and no maritime alternative at all — which is why one strait carries the world's risk premium. Tap a bar for detail. Magenta = elevated risk in the 2026 crisis.
Flat price reacted instantly and forgot fastest — futures are what you can sell at 2 a.m. on a headline. The premium that stayed was administered: war-risk insurance went from ~0.25% of hull value pre-war to quotes near 10% at the peak, settling at 3–8% — that's $3–8 million per Hormuz voyage, payable before the ship moves. Invoice-level evidence showed the war's cost landing almost entirely in the underwriters' line, even on routes that physically bypass the strait, because war risk is priced by whether your destination sits inside the Joint War Committee's listed area — not by the danger your voyage actually runs.
The shadow fleet is a caste system, not a cycle. The authors' own hypothesis — that compliant tonnage returns whenever Urals dips below the price cap — was killed by the data: the G7-linked carriage share sat in a 23–33% band while Urals swung from $56 to $112 and back. Opaque ownership and non-Western insurance are sunk investments that don't un-sink when prices dip. Two-tier tanker markets are architecture.
Spot tanker rates don't move like other asset classes. They gap. Toggle between the two great spikes:
Through the 2026 spike, spot printed near $470,000/day while 12-month charters held around $110,000. That gap is itself a price — of mean-reversion expectations, charterer discipline, owner refusal to sell convexity cheap, or an uncertainty band too wide to price. The working rule: take a disruption thesis seriously not when spot spikes, but when the one-year money agrees.
The tanker fleet is simultaneously the oldest in decades (VLCCs averaging 13.7 years, a fifth of the crude fleet past 20) and carrying its largest orderbook since 2011 (22% of the fleet by April 2026). Not one tanker was demolished in May — sanctioned trades pay geriatric ships too well to die. Age, full yards, and war-trade hold the market tight until roughly 2027; then ships arrive on public schedules anyone can read. Some of the best trades in shipping history came from windows. Nobody reads the closing date. Shipping has spent 200 years proving this.
1. Why did WTI go to −$37.63 while Brent stayed positive the same afternoon?
2. Which premium was "stickiest" during the 2026 Hormuz crisis?
3. A pipeline goes into "apportionment." What is the ladder telling you?