A physical crude price is not a number agreed on the trade date. It is a benchmark averaged over a contractual window, plus a differential for that grade at that place - and futures cover only the first half, which is why a book that is flat against the benchmark is not a flat book.
Summary
A crude oil trade has a price, but that price is almost never a number agreed on the day of the trade. It is a formula: a published benchmark, averaged over an agreed window of dates, plus or minus a differential for that particular grade delivered at that particular place. The futures market gives you a clean instrument for the first half of that formula and nothing at all for the second. A book that is flat against the benchmark is therefore not a flat book. It is a book whose remaining exposure has stopped appearing on the risk report.
This article sets out how a physical crude price is actually assembled, why the part of it that is easy to hedge is often the smaller part of the result, and what a crude trade line has to state before anyone can hedge it, value it, or settle it. It is written for the people who carry the consequences of getting this wrong: traders running physical grade positions, the risk and middle office who report them, and the operations and settlement desks who discover at the invoice stage which of the assumptions held.
If you want the mechanism, start with A crude price is two numbers, and only one of them is quoted. If differentials are already your daily work and you want the part that goes wrong quietly, start with The price is decided after the trade, and a calendar decides it.
A crude price is two numbers, and only one of them is quoted
In physical crude, the flat price and the differential are two separate exposures with two separate causes, and only one of them has a liquid hedge.
When a cargo is bought, what is agreed is rarely a price. What is agreed is a benchmark — Brent, WTI, Dubai or another regional marker — a rule for turning that benchmark into a number over a set of dates, and a differential to be added to or subtracted from it. The benchmark part is the number the market talks about all day and the number the futures curve is built on. The differential part is negotiated between two parties, quoted thinly if at all, and is where the specific identity of your cargo is priced.
Both halves are real money and they move for unrelated reasons. The benchmark moves on everything that moves oil in aggregate. The differential moves on what refiners will pay for that grade, at that place, in that month, relative to the marker — which is a function of refining economics, of which competing barrels are available, and of freight. A week in which the benchmark falls sharply and the differential strengthens sharply is an ordinary week, not a strange one.
The consequence is uncomfortable and worth stating plainly: hedging the benchmark converts a price exposure into a differential exposure; it does not remove the exposure. That is still worth doing — the benchmark leg is usually the larger and more violent of the two — but the residual is not noise. It is the part of the trade that was specific to the barrel you actually bought, and it is now the whole of your remaining position.
The differential is where the barrel's identity is priced
Crude oil is not one substance, and a barrel is not a unit of value. What a refiner pays for is the slate of products the barrel will yield in their particular plant.
Two cargoes can both be crude oil and have very little else in common. One is light and low in sulphur; another is heavy and sour; a third sits somewhere between and carries contaminants that matter to some refineries and not to others. The full description is an assay — a laboratory characterisation of what the crude contains and what it can be made into — and it is the assay, not the word "crude", that determines who wants the cargo.
This is why the differential is not a haggling margin. It is the price of a specific set of physical properties, to a buyer whose plant has a specific configuration, in a market where the alternative barrels available that month have their own properties and their own prices. A refiner comparing your cargo to another is not comparing two prices; they are comparing two yields net of the cost of processing them.
Two things follow, and both of them bite in the middle office. First, a differential is only meaningful alongside the grade, the delivery location and the pricing basis it belongs to — quoting one without the other three is quoting nothing. Second, the differential's risk cannot be understood as a small version of flat-price risk, because it responds to a different set of drivers. A model that treats it as a residual to be smoothed will be calm in exactly the periods when refinery economics are repricing every grade in the region.
The price is decided after the trade, and a calendar decides it
The trade date and the pricing date are different dates, and the pricing date is a range rather than a day.
A physical crude contract fixes a pricing mechanism, not a price. That mechanism names a set of published quotations — commonly a window around the bill of lading date, or the month of loading, or the month of arrival — and takes their average. The parties agree the shape of the window when they agree the trade. The number it produces does not exist until the dates in it have passed.
Several consequences run out of that single fact.
The hedge has to be built from the pricing window, not from the delivery date. A futures position is dated to a contract month. A cargo is priced over a window that may straddle two contract months, may sit at the end of one, and moves if the vessel is late. The mismatch between the two is a real exposure that exists even when the volumes match exactly, and it is not fixed by hedging more.
Two cargoes of the same grade, agreed at the same differential, loading in the same month, can settle at materially different prices — because their windows covered different days and the market moved across them. Neither trader made a mistake. The difference was decided by the contract's calendar.
Many contracts hand one party a choice of window, to be declared before the window opens. That option has value, it belongs to whoever holds it, and it is often recorded nowhere except the contract itself and an email confirming the declaration. An option that is not in the position system is an option nobody is managing.
And an operational delay is a pricing event. When a vessel loads late, the pricing window moves with it. The commercial fact — that a berthing delay has just repriced the cargo — is created in the operations mailbox and typically arrives at the risk report much later, if it arrives as a repricing at all rather than as a quantity update.
The barrel is a claim on a stream, not on a set of molecules
Between the point where crude is bought and the point where it is delivered, its identity is maintained by contract and measurement, not by physics.
Crude moves through systems that mix it. A pipeline that gathers several producers' output delivers a common stream, and what comes out is a blend whose properties are a weighted result of everything that went in. Tanks are shared and turned over. Blending is often deliberate: a cargo can be assembled to meet a specification rather than being found in that state.
So the grade named in your contract is a contractual assertion supported by a measurement, and where the system commingles material, the contract usually carries an adjustment mechanism to settle the difference between the quality a party put in and the quality they took out. That mechanism is the point at which a physical fact — a laboratory result — becomes a money fact.
This is also where the "same" cargo acquires several quantity and quality figures over its life: the figures at load, the figures at discharge, and whatever the contract says shall govern between them. The number that settles is the one the contract nominates, and it may not be the number either party's operations team is watching. Where these two diverge, the dispute is not usually about honesty; it is about which measurement the contract made authoritative, and that question is answerable in advance and very often is not asked in advance.
A flat position is not a flat exposure
Netting barrels across grades produces a number that is arithmetically correct and commercially meaningless.
A book that is long light sweet crude in one region and short heavy sour crude in another can report zero barrels net. It carries a live position in the spread between two grades, in two locations, priced off possibly two different markers, over possibly different windows. Every one of those is a place where the two halves can move apart. The netted line reports none of them.
This is the crude-specific version of a general trap, and it is worth naming precisely because the arithmetic is so persuasive. The unit of exposure in physical crude is not the barrel. It is the combination of grade, location, marker and pricing window — and any aggregation that discards those four fields discards the position while keeping the volume.
The useful test is to ask, of any net figure on a position report, what trade you would have to do to close it. If the answer is a single trade, the netting was probably legitimate. If the answer is "several, at prices we would have to go and discover", then the report has aggregated across a risk rather than netting within one.
Where this leaves the middle and the back office
None of the above is exotic. All of it is ordinary crude trading, and all of it has been ordinary for a long time. What makes it expensive is where the governing facts are kept.
The trading desk's own work — position, mark-to-market, limits — has been systematised for a long time. The steps that decide the numbers in this article mostly have not been. The pricing window and its declaration sit in a contract and an email. The assay sits in a PDF from a laboratory. The loading dates that anchor the window sit with operations. The quality adjustment arrives with the discharge documents. The invoice is checked against all of them in a spreadsheet.
That is not a criticism of anyone's process; it is a description of where the middle and back office of a commodity trading business actually runs. But it has one predictable consequence, and it is the reason the same surprises recur: a fact that decides the price of a cargo reaches the risk report only after somebody has decided it is final. By then, the decision that fact should have informed has already been taken.
The same cargo, four numbers
| Benchmark leg | Grade differential | Quality and quantity adjustment | Freight and location | |
|---|---|---|---|---|
| What sets it | The published marker, averaged over the contract's window | Negotiation, against what refiners will pay for that grade that month | The contract's measurement and adjustment terms, applied to a laboratory result | The freight market and the route actually taken |
| When it becomes final | When the last day of the pricing window has passed | At the trade, unless the contract reopens it | At load or at discharge, whichever the contract nominates | On fixture, adjusted by what the voyage does |
| Where the fact first exists | The price publisher | The trade recap or contract | The inspector's certificate | The charter party and the operations desk |
| Is there a liquid hedge | Yes | Rarely, and never exactly | No | Partly, and separately |
| How it usually goes wrong | The window straddles the hedge's contract month | Carried as a residual rather than as a position | Governed by a clause nobody read until it mattered | Priced on the planned voyage, settled on the actual one |
The bottom two rows are the ones worth sitting with. Three of these four numbers have no clean instrument behind them, and all four have to be right before the fifth number — the one on the invoice — can be right.
What a crude trade line has to state before it can be hedged or settled
| Field | Why it is load-bearing |
|---|---|
| Grade, and the assay it was traded on | The differential is the price of these properties; without them the differential is unattached to anything |
| Delivery basis and location | A differential quoted without its location is not comparable to any other differential |
| The marker the price is built on | Two markers can disagree, and the disagreement is your exposure |
| The pricing window, expressed as dates | The hedge's shape comes from here, not from the delivery date |
| Whether a pricing option exists, who holds it, and whether it has been declared | An undeclared option is an open position held by whoever holds the option |
| Which measurement governs quantity, and which governs quality | This is the clause that decides settlement disputes, and it is decided before the dispute |
| The adjustment mechanism for quality differences | This is where a laboratory result becomes money |
| The event the window is anchored to, and its current expected date | When the vessel moves, the price moves; the position report should show that |
A line carrying all eight fields can be hedged, valued and settled by someone who was not in the room when it was agreed. A line missing any one of them can only be handled by the person who remembers the trade — which works, until they are on holiday and the vessel is late.
Questions people ask about this
Is Brent the price of our cargo?
No. A benchmark is the reference your price is built on, not the price itself. Your cargo's price is the benchmark, averaged according to your contract's rule over your contract's window, plus or minus the differential agreed for your grade at your delivery point. Reading the benchmark and treating it as your price is accurate only in the special case where your differential is zero and your window is today, which is not a case that occurs.
We hedge every cargo on futures — why does our profit and loss still move?
Because futures hedge the benchmark and your cargo is the benchmark plus a differential. When the differential moves, the hedge does not follow it, and the result appears in your profit and loss as a residual with no obvious cause. There is a second, quieter source as well: the hedge is dated to a contract month while the cargo is priced over a window of dates, so even a perfectly sized hedge can be imperfectly timed. Both effects are structural rather than evidence of a mis-executed hedge.
What exactly is a crude differential, and who sets it?
It is the price of the difference between your particular barrel and the marker your contract references — the value the market puts on that grade's properties, at that location, for that delivery period. It is set by negotiation between buyer and seller, informed by what refiners in that region are willing to pay for the yield that grade produces relative to the alternatives available to them. Unlike the benchmark, it is not continuously published in a form you can trade against, which is precisely why it tends to be carried as a residual rather than as a position.
Why is our invoice priced off days we did not choose?
Because the pricing window was agreed at the trade and is anchored to an event, usually the bill of lading date, rather than to a fixed calendar. If loading moved, the window moved with it. This is a contract term working as intended, not an error, but it has a consequence people often have not planned for: an operational delay is also a repricing, and the party who finds out first is the operations desk rather than the risk desk.
Do we need to hedge the differential, and can we?
You are exposed to it whether or not you hedge it, so the first step is to carry it as an explicit position rather than as an unexplained residual. Whether it can be hedged depends on the grade and the region: some differentials have related traded instruments and some have nothing at all, and where a proxy exists it is a proxy — it will track your exposure until the moment the two stop being the same thing, which is usually the moment that matters. Measuring it first is worthwhile regardless of what you conclude about hedging it.
Our position report nets barrels across grades — is that wrong?
It is wrong if the netted number is used as a risk figure. Net barrels is a legitimate operational statistic — it tells you about volume commitments — but a long grade in one location against a short grade in another is a spread position, not a flat book. The practical test is to ask what single trade would close the net figure. If no single trade would close it, the report has aggregated across a risk instead of netting within one.
Why does the quality on the discharge assay change what we are paid?
Because most crude contracts price a specification and then adjust for what was actually delivered. Crude that has been through shared pipelines and tanks is a blend rather than a preserved parcel, so the properties at discharge can legitimately differ from those at load. The contract names which measurement governs and how differences are settled; that clause decides the money. It is worth reading before a cargo is fixed rather than after a difference appears, because it is far easier to agree a rule than to agree an outcome.
Can we run a crude book in spreadsheets?
Many companies do, and the spreadsheet is rarely the thing that fails. What fails is that each cargo's governing facts — the window, the option and whether it was declared, the assay, the measurement clause — live in different places and different formats, so the spreadsheet holds a snapshot of what somebody transcribed rather than a view of what is true now. The scale at which this stops working is not a volume threshold; it is the point at which no single person can still hold every open cargo's terms in their head.
What should we fix first if our crude settlements keep being disputed?
Start with the fields, not the tools. Make the pricing window and its anchoring event a stored, visible property of every open trade, and make the governing measurement and adjustment clause visible on the same line. Those two changes remove the largest category of disputes, which is not disagreement about facts but disagreement about which fact the contract made authoritative. Both are changes to what gets captured and when; whatever holds the result can stay exactly as it is while you make them, and you will find out in the process whether the tool was ever the constraint.
Where this lands in a trading system
Nothing above is an argument for buying software. It is an argument about which facts have to exist, in what form, and how soon. But the article's conclusions do map onto system capabilities, and it is worth being specific about which ones and about where the mapping stops.
The first conclusion is that the benchmark leg and the differential are separate exposures that have to be visible together, alongside the physical cargo they belong to, before either can be managed. That is the job of a CTRM/ETRM system providing complete trading lifecycle management for commodity traders, integrating physical trade, financial hedging, risk control, and settlement in one platform; Time Dynamics' Fusion is one, and the part of it that bears on this article is that complete physical trade lifecycle management and execution and advanced trading management for futures and derivatives markets are held in the same place, which is the condition under which a hedge can be compared to the thing it is hedging.
The second conclusion is that the facts deciding a crude price — the pricing window, the declaration of an option, the assay, the loading dates — usually exist outside that system, in contracts, laboratory reports, counterparty portals and spreadsheets. That is a different problem, and it is the one X-Ray addresses: a non-invasive data processing and analysis platform, whose collection toolkit XDK performs non-invasive automated data collection from databases, Excel files, and web interfaces, and which is built to collect data without disrupting existing systems or workflows. The last property is the relevant one, because the spreadsheets holding your open cargo terms are working, and no plan that begins by replacing them will survive contact with a loading programme.
The third conclusion is that the invoice is the point where all four numbers have to agree, and that checking it by hand is where the cost of everything above is finally paid. Automated financial settlement and payment processing is what a trading system contributes there.
The fourth conclusion is that exposure has to be re-cut — by grade, by marker, by window, by counterparty — faster than a person can rebuild the view. X-Sheet does one-click generation of personalized reports with Excel-like interface.
None of this decides your differential, and none of it has a view on which barrel you should buy. It will not tell you what a refiner will pay for a heavy sour cargo next month, it will not read a contract and tell you the clause is a bad one, it will not choose your pricing option for you, and it cannot make a hedge fit a window that the contract shaped badly. What a system can settle is narrower and duller than the problem in this article: where each of those eight fields is captured, how soon after it exists, and whether the person who needs it can see it without asking someone who was in the room. That is a large part of the gap. It is not the whole of it, and it is worth being clear about which part you are buying.
A note on the evidence in this article
This article contains no statistics, and that is a deliberate choice rather than an omission. The observations behind it — which steps of the trade lifecycle are still run on spreadsheets and email, and which of them carry costs that land directly in money — come from material we are not in a position to publish as a citable figure. A number we cannot source would make the argument look stronger and would make it worth less.
So the argument here is structural instead. Each claim is made from the mechanism, and each one is stated in a form you can check against your own operation rather than take on our authority. The tests proposed above — what single trade would close this net figure, which measurement does this contract make authoritative, on what date did this fact first exist somewhere in the company — are the intended way to use it. If your own book contradicts what is written here, your book is the better evidence.