In short: A term crude price is a published marker plus a differential, and since both sides read the same marker assessment, the differential is the only number the pricing meeting actually decides. The differential compensates for quality against the marker grade, freight to the destination, and the terms attached to the cargo, and each of those can be estimated directly rather than inherited from last month. The same barrel prices differently into different regions because the competing barrels and the product values differ there, so one global differential leaves money on the table in one region and volume unsold in another. The feedback loop runs slowly, since a differential set too high shows up as lost volume over months rather than as an immediate refusal, which is how drift persists.
The monthly pricing meeting rarely spends much time on the marker. It settled where it settled, both sides read the same published assessment, and nobody in the room has a view worth arguing. The number under discussion is the differential: a few dollars either side of that marker, set by grade and by destination region, deciding whether next month's volume moves and what margin comes with it.
It usually gets set by taking last month's number and adjusting for whatever changed. That is a defensible method, and it is also why the differential drifts away from what the barrel is worth, slowly, in whichever direction nobody has been checking.
How a term price is put together
A term price is a formula rather than a number. A published assessment of a marker crude, averaged over a defined pricing period tied to loading or discharge, plus or minus a differential fixed ahead of the month or agreed at the time of the deal.
Each piece does different work. The marker choice is largely conventional and follows the destination region, so a barrel going east and a barrel going west are commonly priced against different markers even when they come out of the same field. The averaging period allocates price risk between the parties and is negotiated once, then left alone for years. The differential is the only term in the whole structure that expresses a view about this barrel, sold to this buyer, into this market, this month.
Everything else in the formula is machinery. The differential is the decision, and for term volumes it gets remade every month, which makes it a repeated decision with a very short cycle and almost no information in any single observation.
What the differential is compensating for
Four things, and keeping them separate is most of the work.
Quality against the marker. The marker crude has an assay and yours has a different one. A refiner converts that difference into a value gap through their own yield economics, which means the gap is computable rather than a matter of opinion, and the calculation is the same one a refiner runs when choosing a slate, covered separately in this series. Density, sulphur, acidity and metals all show up here, and the size of the gap depends on the configuration doing the processing rather than on any universal quality ranking.
Freight to the destination. What the buyer compares is delivered cost. A grade closer to the buyer carries a freight advantage that both parties can compute from published rates, and if you do not price it into the differential the buyer will price it into their offer.
Conditions in the destination market. Local refining margins, competing barrels arriving from other suppliers, regional inventory levels, and the turnaround schedule of the refineries in that region. This is the component that moves fastest and the one most likely to be carried forward stale, because it changes on a different clock from the other three.
The terms wrapped around the barrel. Pricing period, credit terms, destination restrictions, volume flexibility inside the contract, laytime and demurrage provisions. Each of these has value. Left unpriced, that value ends up inside the differential anyway, and once it is in there it is impossible to see what you conceded or what it cost.
Why the same barrel prices differently into different regions
The same grade lands at different values in different regions, because product prices differ, the refineries that can process it differ, and freight differs. A single differential applied everywhere leaves that spread on the table for whoever is best placed to capture it, which is usually the buyer with resale rights.
Destination restrictions in a term contract exist for exactly this reason. They keep the regional spread on the seller's side of the table, and a buyer who negotiates them away has bought something whether or not anyone priced it.
The allocation implication follows directly. Rank destinations by what the barrel actually returns to you, adjusting for freight and for the local marker so the comparison is on a common basis, and the ranking tells you where marginal barrels belong. Term commitments override that ranking most months, and they were made for reasons that had nothing to do with this month's regional spread, which is fine as long as somebody knows what the override costs.
That ranking also moves more than most pricing structures allow for. Refinery turnaround seasons, seasonal demand patterns, and the arbitrage economics of competing barrels reorder it within a quarter. A differential structure that holds a stable regional relationship across a year is not tracking a market that reorders itself every few months.
The feedback loop, and how slowly it runs
Set the differential too wide and the buyer takes the contract minimum, requests a deferral, or declines the optional volume. You now hold a barrel with a limited set of homes, and the alternatives are a spot sale at a worse number or storage.
Set it too narrow and everything lifts, including the optional volumes, and you have no way of knowing how much you left behind. A cargo lifted at a differential thirty cents below the buyer's indifference point looks identical to one lifted at three dollars below it.
That asymmetry is the important part. The wide error announces itself through a phone call, a deferral request or an unlifted cargo. The narrow error is silent and generates no complaint at all. An organisation that learns only from complaints will drift toward the narrow side year after year, and the drift will not be visible in any report it runs.
The counter is to record the counterfactual each month: what the buyer's next best alternative was, and by roughly how much your offer beat it. That is an estimate, and an estimate written down every month becomes a series with enough structure to test.
The volume response also fails to behave like a curve. Term nomination is lumpy, since a buyer takes the minimum, the nominal volume, or the maximum, so what you observe is three levels rather than a smooth response. Learning the shape from that takes years of observations, which is worth knowing before anyone promises an optimised differential from a model.
The netback the buyer is running
The buyer's calculation is straightforward and it is the one that decides whether your cargo moves. For each candidate grade: the value of the product slate it produces at their configuration and their local product prices, minus variable processing cost, minus freight, minus the crude cost. They choose on the difference between grades.
Your differential is therefore competing against their next best barrel rather than against the marker. Several things follow that are useful in a pricing meeting.
The ceiling on your differential is the point at which the buyer is indifferent between your grade and the alternative, and that point moves with the alternative's price rather than with the marker. When a competing grade's differential falls, your ceiling falls too, by an amount that depends on how close the two assays are and on how much conversion capacity the buyer has to exploit the difference.
Their binding constraint changes what your barrel is worth to them specifically. A refinery running against its hydrotreater limit cannot capture the sour discount, so a high sulphur grade is worth less to that buyer than the generic discount implies. A refinery with the coker down for six weeks values a heavy grade much lower for those weeks and higher afterwards.
Understanding the calculation improves your pricing in two directions. It gives you a defensible ceiling instead of a number carried forward, and it tells you which buyer values your particular barrel most, which is an allocation question rather than a pricing one. Selling the same grade at the same differential to a buyer who can use it and one who cannot means the second buyer is either overpaying, in which case they will leave, or setting your price for everyone.
Pricing without seeing the other offers
You never see the competing offers, and you never will. What you do see is a reasonable amount: whether the cargo lifted and at what volume, what the buyer asked for during the negotiation, published assessments for the grades competing with yours, freight rates, arrival and discharge patterns, announced turnarounds, and product cracks in the destination region.
Three ways of using that are worth the effort.
Reconstruct the buyer's netback from public data. Published assays for the competing grades, published product prices in the destination market, published freight, and a workable model of the buyer's configuration will get you their indifference point within a range. A range built from public data beats a number carried forward from last month, and it can be rebuilt every month for the cost of an analyst's afternoon.
Treat the monthly setting as a designed experiment. Vary deliberately in small increments across comparable buyers and comparable regions, record what happened, and keep the record for a year. The aim over that period is to accumulate observations that contain variation, because a differential that moves only when the market moves produces a series with nothing in it to learn from.
Price the terms separately. When a buyer wants a longer pricing period, a wider laycan, or destination flexibility, put a number on each rather than absorbing them into the differential. It keeps differentials comparable across deals and it gives you concessions to trade that are not price.
The limit
Differential setting sits inside a long-term commercial arrangement, and a purely optimisation-driven number ignores what the arrangement itself is worth. A stable term buyer lifts through weak markets, gives you a predictable base for production and marine scheduling, costs less to serve than a rotating set of spot counterparties, and removes the monthly problem of finding a home for the barrel. That is worth real money and none of it appears in a netback comparison.
The failure mode of the fully optimised version is behavioural rather than analytical. A buyer who learns that you extract the last thirty cents every month will build optionality against you: diversify supply, negotiate wider volume flexibility, push for shorter commitments, and move to spot whenever the market allows. Each month's extraction is measurable and the erosion underneath it is not, so the scoreboard favours the behaviour that causes the damage.
There is also a hard limit on how far the analysis can go. One observation per buyer per month, a lumpy nomination response, competitors you cannot see, and a market moving underneath the whole thing means any statistical estimate of the demand response carries wide error bars. A recommendation stated to two decimal places from that estimate is describing a precision it does not have.
And the marker contributes noise of its own. When the assessment window is thin, or when the marker grade's own fundamentals move for reasons unrelated to your barrel, the differential absorbs the difference. A differential series that looks unstable is sometimes telling you about the marker rather than about your pricing.
Reconstruct the netback for your two largest term buyers on last month's cargo, using public product prices, published freight and your own assay, and compare the indifference point you get with the differential you actually set.