Trafigura and the making of a fuel-oil benchmark

Banking on Oil : Part 4

Why would a buyer benefit from making its own fuel more expensive? In February 2017, Trafigura bought 3.6 million barrels of fuel oil in a trading window used to assess a benchmark price. The purchases helped push that price higher. They also benefited financial positions whose exposure exceeded the firm’s physical purchasing exposure. That is the mechanism described by the US Commodity Futures Trading Commission in its settlement order of 17 June 2024. ^[1]

The market in question was specific: high-sulphur fuel oil on the US Gulf Coast, or USGC HSFO. This was not a finding about Brent crude or retail petrol prices. The respondent was Trafigura Trading LLC, the Houston-based entity. Platts, the agency publishing the benchmark, was not charged in this order. ^[1]

Part 1 followed the financing of a cargo, Part 2 mapped the banks behind Trafigura and Part 3 reconstructed Chad’s oil-backed debt to Glencore. This fourth part opens a useful window onto commodity trading. A physical purchase produces information; an assessed price incorporates that information; other contracts use the resulting figure. Following the money therefore requires following the price as it travels between markets.

What the $55 million settlement covers

The CFTC imposed a $55 million civil monetary penalty, but the fuel-oil case was only one of three parts of the settlement. A separate set of findings concerned trading on confidential information misappropriated from a Mexican trading company between 2014 and April 2019. Another concerned confidentiality provisions that impeded voluntary communications by employees and former employees with the regulator between 2017 and 2020. ^[2]

The public order does not allocate the penalty among those three sets of violations. Nor does it identify $55 million as the profit made on the fuel-oil activity. Treating the total penalty as either a fuel-oil-only fine or a measure of the trading gain would be wrong. ^[1]

This was an administrative proceeding resolved by settlement, containing formal CFTC findings. Trafigura consented to the order without admitting or denying its findings or conclusions. It was not a criminal conviction following a trial. The company’s statement, published the same day, reiterated that position. The distinction requires us to attribute the account of the trading to the regulator; it does not reduce an operative regulatory order to an unsubstantiated allegation. ^{[1] [3]}

The price of a particular product in a particular place

A cargo does not have one universal price. Quality, location and delivery terms matter. Physical contracts can use a published benchmark and add or subtract a negotiated amount, known as a differential, to reflect the product or the terms of the deal. The CFTC describes this arrangement in the physical oil markets that use Platts assessments. ^[1]

The basic relationship is straightforward: contract price = agreed benchmark + differential. With the differential unchanged, an extra dollar in the benchmark adds a dollar to the price of each barrel priced on that basis. A contract might refer to a daily value or an average over a specified period. The pricing clause determines how the benchmark affects the invoice. The same change need not reach every fuel buyer in the same way.

The product specification makes the boundaries tangible. In Platts’ June 2026 Americas guide, USGC HSFO represents RMG 380 fuel oil with a maximum sulphur content of 3.5%, on an FOB Houston basis: loaded at the specified terminals. The physical loading window runs from seven to fifteen days after assessment. This current specification shows how narrowly an assessed market is defined; it is not evidence that every clause was identical in February 2017. ^[5]

That specificity is economically useful. A buyer comparing prices first needs to know whether they refer to the same material in the same place. Otherwise, an apparent price gap may simply reflect transport or a difference in quality. A benchmark gives the parties a common reference without eliminating the commercial differences between cargoes.

How Platts constructs a market assessment

For the 2017 market, the CFTC describes Platts’ Market on Close, or MOC, process. Firm bids, offers and transactions within a defined window provided key information for the assessment. With limited exceptions, transactions concluded in the window required actual delivery of fuel oil. A participant was doing more than expressing an opinion: its stated price could lead to a real commercial obligation. ^[1]

A Platts notice dated 17 January 2014, before the events in the case, spells out part of that discipline. An intention to trade had to relate to a bid or offer already published, so that it was available for testing in the market. The price and counterparty had to be explicit. The point was to allow other participants to act on the information rather than leave it as an unverifiable assertion. ^[7]

The general guide dated June 2026 explains that information is gathered throughout the day. The assessment seeks a tradable value at the close, with inputs adjusted to a common specification for product, location and timing. It is not the mechanical average of all reported trades. Selecting and interpreting information is part of the process. ^[4]

This is where two different calculations are often confused. Platts’ daily assessment and a contract’s monthly average are separate objects. The former is produced by a market-assessment methodology. The latter can be an ordinary arithmetic average of daily values that have already been published. In the Trafigura case, that monthly average linked the physical market to the financial positions. ^[1]

An export trade with a larger financial position alongside it

The starting point was an identifiable commercial opportunity. In January 2017, Trafigura’s traders saw an arbitrage between the US Gulf Coast and Singapore: an opportunity to buy in one region and sell in another to exploit the price difference. The CFTC describes an export programme developed and deployed roughly between January and March. ^[1]

During January, the Houston entity entered into contracts to sell approximately 3.5 million barrels to its related Singapore company for delivery in February, March and April 2017. It then needed to acquire fuel to meet those commitments. Higher purchasing prices would make that physical trade less attractive. ^[1]

A long position in derivatives could protect against that risk. A derivative is a financial contract whose value depends on a reference price; a long position benefits when the price to which it is exposed rises. Used to hedge an upcoming purchase, the financial gain can offset the higher cost of the physical material. The CFTC expressly recognises that part of Trafigura’s position served as an economic hedge for anticipated purchases. ^[1]

But the regulator found that the long derivative position was larger than the physical exposure it hedged. The excess was effectively speculative. As a result, a higher benchmark could generate gains on the derivatives exceeding the additional cost of the physical fuel. The buyer’s invoices, examined on their own, would have given an incomplete picture of its economic interest in the price. ^[1]

That is not a reason to treat every hedge as suspect. The export programme supplied a commercial rationale for part of the position. Nor does an additional speculative exposure, by itself, establish manipulation. The further question is what happened in the physical market used to assess the reference price.

Eighty cargoes in the pricing window

From 1 February through the end of the month, Trafigura bid heavily and bought 80 cargoes totalling 3.6 million barrels in the MOC window. The order says trades in that window were generally for 45,000 barrels. Its use of “cargoes” should not be turned into a claim that 80 ocean-going tankers were involved. The public account identifies transactions, not a fleet. ^[1]

The CFTC highlights two departures from the firm’s previous behaviour. Trafigura had never bought so much in the window within a single month. Its near-exclusive use of that window to source such large quantities also differed from its past conduct. The regulator found that the concentrated buying created artificially high benchmark values during February 2017 that did not reflect ordinary supply and demand. ^[1]

Two conspicuous numbers in the order invite a tempting but invalid calculation. The 3.5 million barrels were physical sales contracted in January for three delivery months. The 3.6 million barrels were physical purchases made in the February window. Neither figure is the size of the derivatives position. Subtracting one from the other does not reveal a speculative position of 100,000 barrels. The order does not disclose the size of the excess derivative exposure. ^[1]

The significance of the case lies partly in the fact that the purchases were real. An executed transaction provides evidence of market activity, but that does not automatically make its price representative of an undistorted market. In this case, the regulator considered the physical buying alongside the purchaser’s larger financial exposure. ^[1]

The legal finding was that Trafigura acted with at least reckless disregard for the likely artificial increase in the assessment and the resulting benefit to its derivatives. The CFTC described an extreme departure from the standard of ordinary care. That is a more precise account than claiming that Trafigura admitted a fully documented, intentional scheme. It made no such admission in the settlement. ^[1]

Nineteen daily values, one monthly reference

The financial positions described in the order settled by reference to the average of the daily Platts benchmark values over the 19 trading days in February 2017. A physical purchase could therefore affect a daily assessment, which then entered a monthly average used to value financial contracts. ^[1]

A currently documented ICE contract illustrates that link. The Fuel Oil Outright – USGC HSFO (Platts) Future, symbol RBO, represents 1,000 barrels and is cash-settled. Its final settlement uses the average of the daily “Mid” quotations in the Houston section of Platts US Marketscan during the determination period. The specification consulted on 7 September 2026 does not identify the particular contracts Trafigura held in 2017. ^[6]

A simple sensitivity calculation shows what one day contributes, without inventing a historical price series. In an equally weighted average of 19 values, adding $1 per barrel to one value, with the other eighteen unchanged, raises the average by $1/19, or approximately 5.26 US cents per barrel. Adding a dollar to all nineteen values raises the average by a dollar.

A monthly average thus dilutes an isolated movement but does not cancel a repeated influence. This says nothing about the actual size of the distortion in 2017. The CFTC does not publish the nineteen counterfactual prices that would have prevailed without the conduct, or an estimate of their difference from the observed values. Drawing a historical “fair price” line from this public record would mean inventing data.

The gain on a financial position also depends on its size and entry price. The result of the whole commercial operation requires the physical purchasing cost and other expenses as well. Without those inputs, the 3.6 million barrels of purchases cannot be converted into a credible estimate of net profit attributable to the manipulation. ^[1]

Publishing a price when the visible market is thin

The case raises a broader measurement problem: what happens when only a limited part of the physical market supplies observable information? In its 2 October 2012 summary of IOSCO’s principles for oil price-reporting agencies, the Financial Stability Board noted that data submission was voluntary. It also recognised that overly burdensome requirements could discourage contributions. This is the framework described in that 2012 document, not a comprehensive statement of every reporting obligation in force in 2026. ^[8]

Platts’ June 2026 methodology does not prescribe a minimum volume of transaction data for an assessment. Where one company supplies more than half the available information, its bids or offers must be executable by other potential MOC participants. That is neither a market-share measure nor an automatic prohibition. The CFTC order does not disclose Trafigura’s share of all the information underlying the February 2017 assessments. ^[4]

There is a genuine trade-off here. Refusing to assess a price whenever trading becomes thin could deprive contracts of a usable reference. Continuing to publish requires confidence that the available observations still support a reasonable assessment. The number of transactions, their verifiability and the ability of other participants to respond answer different questions. None automatically substitutes for the others.

Actual execution is therefore a useful check, but an incomplete one. Detecting the mechanism in this case also requires connecting physical interventions with the financial positions that benefit from the published benchmark. That is an implication of the CFTC’s account. It is not evidence that Platts knew the entire derivatives position or participated in the conduct sanctioned by the regulator. ^[1]

Public responses and the limits of the record

Trafigura’s June 2024 statement says it strengthened market-integrity procedures, communications controls and compliance testing after the period in question. That is the company’s public response, not independent verification of the effectiveness of its improvements. ^[3]

There were also objections inside the CFTC. Summer Mersinger and Caroline Pham, commissioners at the time, criticised the new interpretation applied to confidentiality provisions. Their objections concerned that separate part of the case, not a rebuttal of the fuel-oil manipulation analysis. Pham expressly commended the enforcement work against manipulation. Their statements cannot fairly be presented as a rejection of the entire case. ^{[9] [10]}

The main unresolved questions are quantitative. In the public documents reviewed for this article, we found neither the exact excess derivative exposure, nor net profit attributed to the manipulation, nor losses allocated among customers. The order sets out the regulator’s conclusion and the mechanism behind it. It does not release the full investigative record. A distorted contractual reference can reach firms that never participated in the initial purchases; the effect on any one of them depends on its contracts and hedges. ^[1]

The documented risk arises from the mismatch between the exposures: physical buying can affect a benchmark to which a larger financial position is tied. The case explains why purchases, pricing periods and derivatives need to be examined together. An invoice shows what the buyer paid for the fuel. On its own, it does not show how the same movement in price affected the rest of the trade. ^[1]

Documentary investigation as of 7 September 2026. Market events: February 2017. Regulatory order and public responses: 17 June 2024. Current methodology documents: June 2026; ICE specification consulted as of the research date. No interviews or directly obtained responses are claimed.

Sources and documentary notes

Sources reviewed as of 7 September 2026. Page links refer to PDF pages, which match printed pagination in the cited documents. The 2026 materials describe current rules, not a complete archive of the 2017 methodology. Company representations and access limitations are identified.

1. CFTC : In the Matter of Trafigura Trading LLC : Docket No. 24-08. Administrative order of 17 June 2024, entered without admission or denial of its findings. P. 1: settlement status; pp. 3–4: benchmark, MOC and usual lot size; pp. 5–6: physical sales, February purchases, derivative exposure and 19 trading days; pp. 8–9: legal standard; p. 11: total civil penalty. No disclosed amount for excess derivative exposure or attributable net profit. Printed and PDF page numbers match.
2. CFTC : CFTC Orders Trafigura to Pay $55 Million for Fraud, Manipulation and Impeding Communications with the CFTC : Release 8921-24. Release dated 17 June 2024. Confirms the three parts of the settlement and aggregate penalty; the order supplies the transaction and exposure details. The other conduct covers 2014–April 2019 and 2017–2020.
3. Trafigura : Statement re Civil Settlement with US CFTC. Company statement published on 17 June 2024. Neither admission nor denial; compliance improvements are company representations, not independently verified here. This is a published statement, not a response obtained by l0g.
4. S&P Global / Platts : Platts Assessments Methodology Guide. Version labelled “Latest update: June 2026”. P. 4: all-day collection and assessment at the close; p. 10: no minimum transaction-data threshold, concentrated submissions and normalisation; pp. 11–12: hierarchy and judgment. Published methodology, not an audit of its application in 2017 or 2026. Relevant PDF pages, including the process diagram on p. 4, were inspected.
5. S&P Global / Platts : Specifications Guide : Americas Refined Oil Products. June 2026 version, p. 51: FOB Houston, seven-to-fifteen-day loading window, RMG 380 and 3.5% maximum sulphur. Used as a current specification, not as a February 2017 archive. Illustrative prices elsewhere in the guide are not treated as an observed price series.
6. ICE : Fuel Oil Outright – USGC HSFO (Platts) Future : contract specifications. Undated specification consulted on 7 September 2026. RBO contract, 1,000 barrels, cash-settled; Final Settlement uses an average of Platts US Marketscan “Mid” quotations. Daily and final settlement have different definitions. This specification is not identified as the exact position held by Trafigura in 2017.
7. Platts : Transactional interest in bids, offers in Platts MOC process. Notice dated 17 January 2014, before the events. Published bids and offers, firmness, explicit price and counterparty. Establishes these stated requirements at that date, not how every 2017 transaction was reviewed.
8. Financial Stability Board : Principles for Oil Price Reporting Agencies. Institutional summary of IOSCO principles, dated 2 October 2012: reliability, detecting abuse and voluntary data contributions. The FSB summary was consulted; its linked original IOSCO PDF was not retrieved. No comprehensive claim about reporting law in 2026 is inferred.
9. CFTC : Summer K. Mersinger : Concurring Statement Regarding Settlement With Trafigura Trading LLC. Statement of 17 June 2024 by a commissioner then in office. Objections concern the interpretation applied to confidentiality clauses, not a rebuttal of the fuel-oil findings. No claim is made about her current role.
10. CFTC : Caroline D. Pham : Statement Regarding Settlement Order with Trafigura Trading LLC. Statement of 17 June 2024. Commends enforcement against manipulation and criticises the additional confidentiality charge. Used to identify the scope of the disagreement, not to imply rejection of the benchmark case.