Trading Houses
A handful of private firms move a large share of the world’s oil, metal, and grain, earn billions doing it, and almost no one can explain exactly how. The answer is part logistics, part information, part balance sheet, and partly access to officials in commodity producing countries.
The commodity trading houses are the middlemen between producers and consumers of physical raw materials. Vitol, Glencore, Trafigura, Gunvor, and Mercuria dominate energy and metals; Cargill, ADM, Bunge, and Louis Dreyfus, the so-called ABCD firms, dominate agriculture; and Japan’s sogo shosha are a category of their own. They own ships, tanks, terminals, refineries, and mines, and they buy a cargo in one place and sell it in another. That sounds simple and low-margin, yet in the 2022 energy crisis several of them earned the largest profits in their history, Vitol alone clearing on the order of fifteen billion dollars. Understanding how a business of pennies-per-barrel middlemen produces numbers like that is the whole puzzle.
Where they sit, and why
The energy and metals houses cluster in Switzerland, above all Geneva and the low-tax canton of Zug, with Singapore as the Asian hub and London and Houston as trading floors. The Swiss pull is a stack of advantages: very low corporate tax (Zug and Geneva long offered special rulings for trading companies), political neutrality and stability for firms that deal with every regime on earth, a deep cluster of trade-finance banks, inspectors, and lawyers, and a time zone that bridges Asian and American hours. Once the talent and the banks were there, everyone else had to be there too. The cluster was seeded by one man: Marc Richset up in Zug in 1974, and the firm he built became Glencore, while alumni and imitators spread across Lake Geneva. Japan’s houses are different animals. The sogo shosha, Mitsubishi, Mitsui, Itochu, Marubeni, and Sumitomo, are sprawling conglomerates that grew out of the drive to secure raw materials for a resource-poor island, and they combine trading with equity stakes in the mines and projects themselves; Warren Buffett’s Berkshire famously bought into all five in 2020. The agricultural ABCD firms, by contrast, sit close to the farms, in the US Midwest and the Netherlands.
How they actually make money
There is no single secret; there is a bundle, and the firms that win own all of it at once. The textbook layer is arbitrage in three dimensions: space (buy where a commodity is cheap, ship it where it is dear), time (buy when it is in surplus, store it, sell it forward when the curve pays you to wait), and form (blend grades, qualities, or specifications so a cheap input meets an expensive specification). On its own that arb is thin and quickly competed away. What makes it durable is everything wrapped around it: information from seeing real physical flows that no screen shows; optionality embedded in owning the ships, tanks, and terminals, so they can always find a home or a delay for a cargo; an enormous balance sheet funded by revolving bank credit lines secured against the inventory itself, which lets them move volumes a fund could never finance; and increasingly asset ownership, especially Glencore, which is as much a miner as a trader. Add volatility: when markets dislocate, arbs blow wide open and only the firm with the logistics and the credit to act can capture them, which is exactly why 2020 and 2022 were record years.
Is it insider trading? And is that even a thing here?
The houses say their edge is information, and that raises the obvious suspicion. US securities law, which bars trading shares on material non-public information, mostly does not reach physical commodities or the bulk of commodity futures. Trading on your own superior read of supply and demand, your own cargoes, your own flow data, is the business, and it is generally legal. What is not legal is fraud and market manipulation: spoofing, cornering, and feeding false prices into a benchmark. The US Commodity Futures Trading Commission gained broad anti-manipulation and anti-fraud authority after 2010, and the EU goes further, its Market Abuse Regulation and the energy-specific REMIT rules do prohibit insider dealing in commodity derivatives. So the information edge is mostly lawful; the line the houses cannot cross legally is manipulation, trading with an intent to move prices away from where they would otherwise have moved.
The benchmark squeeze, proven
There is a specific worry that a house could use physical control over a thin slice of a market to move an assessed benchmark, then profit on a much larger derivative position tied to it. This is not just theory. Since 2020 close to a billion dollars in fines has been paid to US and Swiss authorities by trade houses, with benchmark manipulation among the conduct at issue. So the squeeze is real and has been punished. It is the abuse at the edge of the model, though, not the everyday engine, which remains the legal bundle of logistics, information, and balance sheet. How an assessed price can be pushed at all is the subject of the price-discovery theme.
The bribery era, and whether it was ever legal
The darkest answer to “how do they win deals” has now been settled in court. Since 2020 a wave of US and Swiss enforcement has produced guilty pleas across the industry. Since 2019 officials in commodity producing nations, Brazil, Ecuador, Mexico, Venezuela, Nigeria, Congo, Ivory Coast, the Democratic Republic of Congo, and more, have been identified in court cases as having received bribes from trade houses. To the question of whether this is legal where the houses are based: no. Bribing foreign officials is a crime under the US Foreign Corrupt Practices Act (which reaches them through the dollar), the UK Bribery Act, and Switzerland’s own anti-corruption law. For a long stretch enforcement was lax and the risk was quietly treated as a cost of doing business; that era is closing, at least in the West. China is the asymmetry: it criminalised bribing foreign officials in 2011 (Article 164 of its Criminal Law, up to ten years) and ratified the UN anti-corruption convention in 2005, yet has brought virtually no prosecutions for what its companies do abroad. The law is widely called a paper tiger, so a Chinese state trader bribing an official in a producing country faces real exposure mainly from the host country or, through a dollar nexus, the United States, rarely from Beijing.
It is worth naming the root of the practice. Marc Rich, the founder of the Zug firm that became Glencore, built his fortune trading with places others would not, embargoed Iran, apartheid South Africa, Cuba. He was indicted in the United States in 1983, fled to Switzerland, ran the business in exile, and was controversially pardoned by President Clinton in 2001. The original edge of the modern trading house was a willingness to operate in the legal and political gray zones that banks and oil majors avoided. Much of the last decade has been the industry being forced, expensively, out of that zone.
Why don’t hedge funds just copy them?
If the trades are so good, why can a hedge fund not simply run them? Because the edge is not a trade you can put on through a screen; it is physical infrastructure and relationships. A fund trades paper. It cannot take delivery of two million barrels of crude on a supertanker and find a refinery to take them, cannot blend off-spec fuel into spec, cannot draw on a fifty-billion-dollar revolving credit line secured against tankage it owns, and does not have an office and a thirty-year relationship in the producing country. The money sits in the physical optionality and the flow information, and both come only from actually owning and moving the material. Pure paper commodity funds have a famously poor long-run record precisely because, stripped of the physical leg, they are taking directional bets against the very firms with better information, which is part of why a nine-figure blowup like Amaranth in natural gas is the genre’s recurring story while the houses compound.
Why so few, and why nobody seems to leave
The premise that no new houses appear is only half right. The barriers to entry are immense: a startup cannot conjure the bank credit lines, the global terminals, the producer relationships, or the back-office risk machinery that took decades to build. And yet senior traders do leave and build, when they can assemble capital and a banking syndicate behind a proven name. Mercuria was founded in 2004 by two former bank-and-trading-house traders, Gunvor around 2000, and Trafigura itself was spun up in 1993 by Marc Rich alumni. So the cluster does renew, just rarely, because the thing a departing star must rebuild is not a strategy but an entire institution of credit, assets, and trust. That is the real reason you do not see a steady stream of one-person breakaways: the platform, not the person, is most of the edge.
Why is there no Chinese house?
This premise is also half wrong: China has enormous trading capacity, it just sits inside state-owned, vertically integrated champions rather than independent merchant houses. Unipec (Sinopec’s arm) and Chinaoil (PetroChina) are among the largest crude traders on earth; COFCO International is China’s deliberate “Chinese ABCD,” assembled by buying Nidera and Noble Agri and, tellingly, run from Geneva and Singapore rather than China; and Sinochem, China Minmetals, and CMOC (which owns copper-cobalt mines in the Democratic Republic of Congo) cover chemicals and metals. So the question is really why there is no private, Geneva-styleChinese house, and the answer is structural. A global house must move capital freely and finance trades in dollars worldwide, which China’s capital controls and limited currency convertibility make almost impossible for a private firm; the state-owned traders manage it only because the state grants them dollar access. The mandate is supply security for the nation, not arbitrage profit, which naturally produces state enterprises rather than profit-maximising merchants. The bank credit, counterparty trust, and English-law plumbing a house runs on are clustered in the West, so China bought into that cluster rather than rebuilding it at home. And Beijing has little appetite for a hugely profitable private trader sitting on strategic flows outside its control; the one serious private attempt, the copper trader Maike Group, nearly collapsed in a 2022 liquidity crisis, which rather made the state’s case. China’s trading houses exist, in other words, but they are state champions that often trade out of Singapore, not an independent merchant class.
The seat or the skill?
Which brings the multimillion-dollar pay packages into focus. Is a star trader at one of these firms worth the money for personal genius, or for the seat: the flow information, the balance sheet, the assets, the official connections, the credit lines that surround the chair? Overwhelmingly the seat. Take the trader out of the platform and most of the alpha stays with the platform, which is why individuals so rarely leave and reproduce the results alone. The pay is the firm sharing the rents of the platform with the people who operate it, and the structure reinforces it: Vitol, Trafigura, Gunvor, and Mercuria are privately owned by their own senior staff, so the top traders are already partners with equity, richly paid and structurally bound to the seat rather than tempted away from it. The handful who did walk out and build a Mercuria are the exceptions that reveal the rule, that what looks like individual brilliance is mostly a chair wired into a machine almost no one can rebuild from scratch.