How Commodity Trading Houses Work: Physical Trading, Futures Hedging, Trade Finance and Enforcement

A commodity trading house is a firm that buys physical oil, metals, grain or gas in one place, moves, stores, blends or finances it, and sells it somewhere else, while hedging the price of what it holds with futures and swaps. The largest, such as Vitol, Glencore and Trafigura, handle volumes comparable to whole countries, yet their profit margins are thin: they earn the difference between what a barrel or a tonne is worth at different places, times and qualities, not the price level itself.

In short: a trading house is a logistics and financing business with a hedged price book. When it buys a cargo it usually sells futures against it the same day, so a fall in the flat price is offset by a gain on the hedge. What remains is basis risk: the spread between its physical cargo and the futures benchmark. That spread is the product. Margins are around 1% of revenue, so the business depends on scale, cheap short-term bank credit, cash to meet margin calls, and tight risk limits. The same structure, a physical book next to a derivatives book that references price benchmarks, is also where the industry’s biggest enforcement cases came from.

This piece explains the model the way the firms themselves describe it in their financial statements. It is a description of how professionals operate, not a strategy to copy.

A common misconception: “trading houses bet on commodity prices”

Trading houses do take some price views, and they disclose risk limits that allow for it. But their own reports describe a business built to avoid outright price exposure. Trafigura’s 2025 Annual Report (financial year ended 30 September 2025) states that its “general policy is to hedge benchmark flat price exposure related to unsold inventories and physical transactions on a deal-by-deal basis,” and that the group “hedges a large majority of price risks arising from its activities.” Its inventory note says company policy is that inventory (other than supplies) “has either been pre-sold or hedged.”

Glencore makes the same point through an accounting choice. Its 2025 Preliminary Results define readily marketable inventories (RMI) as inventories readily convertible into cash because of liquid markets and “the fact that price risk is primarily covered either by a forward physical sale or hedge transaction.” At 31 December 2025, Glencore classed $28.2 billion of inventory as RMI and deducts it when it reports net debt.

So the right mental model is not a hedge fund with ships. It is closer to a bank’s market-making desk, applied to physical goods: inventory is held because the business requires it, and the price risk on that inventory is laid off in the futures market as quickly as possible. The basic mechanics of a short hedge are in how hedging with futures works.

The three biggest independents, by their own numbers

Firm	Period	Revenue	Profit measure	Volumes
Vitol	Calendar 2025	Turnover $343 billion (2024: $331 billion)	Not published	8 million barrels a day of crude and products (2024: 7.2); 23 million tonnes of LNG (2024: 18)
Trafigura	Year to 30 September 2025	$240.3 billion	Net profit $2.67 billion; underlying EBITDA $8.17 billion	358 million tonnes of oil, products, gas and LNG, about 7.6 million barrels a day
Glencore	Calendar 2025	$247.5 billion (whole group)	Marketing adjusted EBIT $2.9 billion; group adjusted EBITDA $13.5 billion	Mining plus marketing

Sources: Vitol’s “2025 volumes and review”; Trafigura’s 2025 Annual Report and results release of 9 December 2025; Glencore’s 2025 Preliminary Results (18 February 2026). Vitol is privately owned by its employees and does not publish profit figures in its review, so none is given here.

Two ratios show the economics. Trafigura’s net profit was about 1.1% of revenue ($2,666 million / $240,268 million). Glencore’s marketing EBIT was about 1.2% of its group revenue, although that understates the marketing margin because Glencore’s revenue also includes its mines. Glencore guides investors to a long-term, “through the cycle” marketing EBIT range of $2.3 billion to $3.5 billion a year, which tells you management expects a business that rarely loses money but also rarely shoots the lights out.

Glencore is the odd one of the three: it is a listed miner with a trading arm that sells its own production and third-party volumes. Vitol and Trafigura are employee-owned. Trafigura’s report links that ownership to risk culture: senior managers have “their own capital at risk.”

How a hedged cargo actually makes money

The basis is the product

A trader who buys a cargo of crude in the US Gulf and sells it to a refinery in Europe a month later faces two separate price questions:

Flat price: where will the global benchmark be in a month? This is what futures hedge.
Basis: what will this particular grade, at this location, on these dates, be worth relative to the benchmark? This is where the trader’s information, logistics and customer relationships pay off.

Buying at a discount to the benchmark and selling at a premium (or a smaller discount) earns the change in basis. Everything else, including freight, storage, insurance, inspection and interest on the money tied up in the cargo, is a cost the basis must cover. (For a refresher on the term, see the basis glossary entry.)

Worked example: one cargo, three outcomes

Assumptions, all illustrative:

A trading house buys 1,000,000 barrels of crude at $69.60 a barrel, when the relevant futures contract trades at $70.40. Entry basis: minus $0.80.
It sells 1,000 NYMEX Light Sweet Crude Oil futures against the cargo (each contract is 1,000 barrels, so a one-cent move is worth $10,000 across the position).
It expects to deliver to a buyer in 30 days at $0.10 over the futures price.
Logistics (freight, storage, insurance, inspection): $0.18 a barrel, or $180,000.
Trade finance on the $69.6 million cargo at 5.5% for 30 days (actual/360): $319,000.

Outcome	Physical P&L	Futures P&L	Basis gain	Net after $499,000 of costs	Unhedged net
Benchmark falls $6 to $64.40, basis as planned	-$5,100,000	+$6,000,000	+$900,000	+$401,000	-$5,599,000
Benchmark rises $6 to $76.40, basis as planned	+$6,900,000	-$6,000,000	+$900,000	+$401,000	+$6,401,000
Benchmark unchanged, basis only improves to -$0.60	+$200,000	$0	+$200,000	-$299,000	-$299,000

Three lessons sit in that table.

First, the hedge makes the result independent of the flat price. Whether crude falls or rises $6, the hedged trader earns the same $401,000: the $0.90 improvement in basis, minus costs. The unhedged trader’s result swings by $12 million on the same cargo.

Second, the break-even is the basis, not the price. Costs here are $0.499 a barrel, so the basis has to improve by about 50 cents just to pay for freight and money. In the third row, the flat price did nothing and the trader still lost money because the basis moved less than planned. That is basis risk, and Trafigura’s annual report names it explicitly: when hedging instruments do not match the physical exposure, “the Group remains exposed to a price risk referred to as basis risk.”

Third, the hedge creates a cash problem even when it works. In the second row, futures are marked to market every day, so as crude climbs $6 over the month the trader pays $6 million in variation margin to its clearing broker before it has been paid for the cargo. The offsetting physical gain arrives only when the buyer pays. This is why trading houses keep large committed credit lines. Trafigura’s report says it uses corporate credit facilities for needs “such as margin calls or bridge financing,” and that at 30 September 2025 it had $14.6 billion of immediately available cash in liquidity funds and unused committed facilities. Its receivables note also shows $7.1 billion of “broker balances.”

Physical delivery is the anchor

Most hedges are closed out before expiry, but the possibility of delivery is what keeps futures prices tied to physical value. Trading houses are among the few participants that actually make and take delivery, and they use exchange mechanisms such as the exchange for physical (EFP) to swap a futures position for a physical one at an agreed spread. The plumbing is in physical vs cash settlement and physical delivery in depth; the exchange rules governing EFPs are in block trades and EFRPs.

Trade finance: running on borrowed working capital

A cargo of 1 million barrels at $70 ties up $70 million for weeks. A trading house moving 7 to 8 million barrels a day cannot fund that from equity. It borrows short term against the goods.

Trafigura’s 2025 report describes three pillars of funding: transactional facilities, securitisation, and corporate credit facilities. Day-to-day trading is financed mainly through “uncommitted, self-liquidating trade finance facilities.” Self-liquidating means the loan is repaid from the sale of the specific cargo it financed: the bank lends against a letter of credit, a bill of lading or a receivable, and the buyer’s payment retires the loan. Uncommitted means the bank can decline to roll it, which is why the corporate facilities sit behind it as a backstop.

The numbers show how central banks are. In its 2025 financial year Trafigura refinanced its Asian revolving credit and term loan facilities at $3.2 billion with 38 banks, then after year end refinanced again at $3.4 billion with 43 banks. It reported $5.3 billion of trade receivables discounted (sold to banks) at year end. Glencore reported $13.3 billion of committed revolving credit facilities ($9.385 billion plus $3.9 billion) at 31 December 2025, which it notes contain no financial covenants, rating triggers or material adverse change clauses.

The financing model has a direct link to hedging. Banks lend against inventory far more readily when the price risk on that inventory is hedged, because the collateral value is then far less volatile. Hedging is not just risk management for a trading house; it is a condition of its funding.

Risk limits: how the house keeps the book small

Both listed disclosures report value at risk (VaR), a statistical estimate of the loss on a normal bad day.

	Trafigura (FY to 30 September 2025)	Glencore (calendar 2025)
Measure	1-day, 95% confidence, historical simulation	1-day, 95% confidence, Monte Carlo
Average VaR	$48.7 million (prior year $55.0 million)	$67 million (prior year $53 million)
Limit or target	Below 1% of group equity	$200 million group limit
Average as share of limit	0.30% of equity, against a 1% target (about $162 million)	34% of the limit

Glencore reported a 2025 high of $118 million and a low of $42 million, and no limit breaches. Trafigura’s model covers “more than a thousand market risk factors,” weights recent data more heavily, and is backtested. Both firms also list the limits of VaR: it says nothing about losses over longer horizons or about the cost of liquidating large positions quickly.

Put those VaR figures next to equity. Trafigura’s average 1-day 95% VaR is about 0.3% of its $16.2 billion of equity. A firm that bet heavily on flat price would show a far larger number. The low VaR is the quantitative signature of the hedged model described above.

When the model goes wrong

Hedging removes price risk. It does not remove the other risks of a physical business, and the past few years produced public examples of each.

Fraud in the physical chain. On 9 February 2023 Trafigura disclosed a “systematic fraud” involving containerised nickel in transit during 2022: containers inspected on arrival “were found not to contain nickel.” It recorded a $577 million charge in the first half of 2023. A perfect futures hedge on nickel that does not exist is worthless.
Credit losses. Trafigura’s 2025 receivables note carries a $708.9 million provision on its Mongolian petroleum products supply business, which its chief executive’s letter links to “serious misconduct by individuals” there.
Liquidity. As the worked example shows, rising prices drain cash through margin calls before the physical side pays. A trading house that cannot fund its margin has to cut its hedges or its book at the worst moment.

The enforcement record, 2020-2024

The trading model gives a firm two books that touch the same prices: a physical book and a derivatives book. Several resolutions in 2020-2024 showed how that combination, plus access to state-owned oil companies, was abused.

Date	Firm	Authority and outcome	Conduct described by the authority
December 3, 2020	Vitol	DOJ deferred prosecution agreement, over $135 million combined with Brazil; CFTC order of $95.7 million (partly offset by the DOJ penalty)	Bribes to officials at Petrobras, Petroecuador and Pemex; CFTC also found attempted manipulation of two S&P Global Platts fuel oil benchmarks
May 24, 2022	Glencore	Guilty pleas in the US to foreign bribery and market manipulation, over $1.1 billion to resolve the DOJ cases; CFTC order of $1.186 billion (with offsets)	Bribes in Brazil, Cameroon, Nigeria and Venezuela, among others; manipulation of US Platts fuel oil benchmarks and related futures and swaps, conduct from as early as 2007 to at least 2018
November 3, 2022	Glencore Energy UK	UK Serious Fraud Office: about £281 million (penalty, confiscation and costs)	$29 million of bribes for oil access in five African countries
December 14, 2023	Freepoint Commodities	DOJ deferred prosecution agreement, over $98 million	Bribes to Petrobras officials for confidential pricing information
March 1, 2024	Gunvor	Guilty plea, over $661 million	Bribes to Ecuadorian officials to win Petroecuador business
March 28, 2024	Trafigura	Guilty plea, over $126 million ($80.4 million fine plus $46.5 million forfeiture)	Bribes to Petrobras officials, roughly 2003-2014
June 17, 2024	Trafigura Trading LLC	CFTC order, $55 million	Trading on misappropriated Mexican pricing information; manipulating the Platts US Gulf Coast high-sulfur fuel oil benchmark in February 2017; employment agreements that impeded whistleblowing

Sources: DOJ, CFTC and SFO press releases of the dates shown; Glencore and Trafigura annual disclosures. Glencore’s 2025 Preliminary Results add that Swiss prosecutors closed their case in August 2024 with a CHF 2 million fine and a $150 million compensation claim, that the Dutch case was dismissed, and that the US Department of Justice terminated Glencore’s monitorships in March 2025. Trafigura’s 2025 Annual Report says a Swiss court fined its former parent CHF 3 million, plus disgorgement, over payments in 2009-2011, and that the decision is under appeal.

Why the manipulation cases matter to futures users

The bribery cases are about access to supply. The manipulation cases are about benchmarks. Much physical oil is priced off assessments published by S&P Global Platts, which are built partly from bids, offers and trades in a short end-of-day window. Swaps and some futures settle against those assessments.

A firm holding a large derivatives position that settles against a benchmark has an incentive to move that benchmark with its physical trading. The CFTC’s 2024 order describes exactly this pattern at Trafigura: heavy bidding and buying during the assessment window in February 2017 “created artificially high benchmark values,” which “benefited Trafigura’s long derivatives position.” For anyone hedging with contracts that settle on a price assessment, this is the reason exchanges and regulators police the link between physical trades and the benchmarks behind them. The wider enforcement picture is in spoofing and manipulation cases.

What can go wrong for the trading house, summarised

Basis moves against the cargo. The hedge protects flat price, not location, quality or timing spreads.
Margin calls outrun financing. Futures losses are cash today; physical gains are cash later.
Bank lines are uncommitted. Trade finance can shrink when banks get nervous, often when prices are most volatile.
Counterparty and fraud risk. The physical goods, documents and buyers must be real and solvent.
Legal and conduct risk. Bribery and benchmark manipulation have produced fines in the hundreds of millions to over a billion dollars, guilty pleas and monitorships.

The short version

Commodity trading houses buy, move, store and sell physical commodities and hedge the price of almost everything they hold, so their profit comes from basis, logistics and financing rather than from price direction. Their own disclosures show margins of about 1% of revenue, VaR of a fraction of a percent of equity, tens of billions of dollars of hedged inventory, and dozens of banks providing self-liquidating trade finance. Futures are what make the model financeable, and margin calls are what make liquidity its main stress point. The 2020-2024 cases against Vitol, Glencore, Freepoint, Gunvor and Trafigura show where the conduct risk sits: in access to state-owned suppliers and in the link between physical trading and the benchmarks that derivatives settle against.