How the LME Cancelled $12 Billion of Nickel Trades in 2022

At 5:42 in the morning London time on 8 March 2022, a nickel contract for delivery in three months printed at $101,365 a tonne on the London Metal Exchange's electronic platform. Ten weeks earlier the same contract had traded near $20,000. Most of the last leg — a doubling — had taken about an hour, in a session when the Ring was dark, the exchange's London staff were mostly asleep, and the buying came from a Chinese stainless steel producer trying to close out a short position it could not deliver against.

By 08:15 the exchange had suspended the market. By late afternoon it had done something no major venue had done at that scale in living memory: it reached back into the morning, voided every nickel trade executed since midnight, and reset the market to the previous evening's close. Around 9,000 trades worth roughly $12 billion in aggregate value simply stopped having happened.

The exchange that did this was 145 years old, had been trading metal since 1877, and had been owned since 2012 by Hong Kong Exchanges and Clearing, which had paid £1.388 billion for it. What the morning of 8 March exposed was not a rogue trader or a fraud. It was a clearing house arithmetic problem that the exchange's own risk controls had failed to see building for three months.

The Hedge That Stopped Being a Hedge

Xiang Guangda built Tsingshan Holding Group into the largest nickel producer in the world out of a car-parts workshop in Wenzhou. Chinese commodity traders called him Big Shot. His company's output was nickel pig iron and, later, nickel matte — cheap material destined for stainless steel mills, produced at enormous scale in Indonesia.

Rising nickel prices are good for a nickel producer, which makes the position Tsingshan had accumulated by early 2022 worth examining. The company was short. Over the preceding months it had sold LME nickel futures against its own future production, a textbook producer hedge: if the price fell, gains on the short would offset thinner margins on the metal.

One detail made the hedge structurally unsound. LME nickel contracts settle in class 1 refined nickel of 99.8 per cent purity — briquettes, cathodes, pellets. Nickel pig iron is not class 1, and neither was most of what Tsingshan made. A producer holding a short in a contract it cannot physically deliver into is not hedged; it is betting that it can buy the position back at a price it likes. Estimates of the size of that bet ranged from 100,000 to 200,000 tonnes on exchange, with press reports putting Tsingshan's total short across LME and over-the-counter positions nearer 300,000 tonnes.

Russia invaded Ukraine on 24 February 2022. Russia supplied roughly a fifth of the world's class 1 nickel, most of it from Norilsk Nickel, and the deliverable grade was precisely the grade the market now feared would be sanctioned out of reach. Nickel had opened February at $22,764 a tonne. It touched $25,575 on the day of the invasion, an eleven-year high. LME warehouse stocks of class 1 metal had been falling for a year.

A short position in a scarce deliverable, held by a producer who cannot deliver, against a supply shock. The squeeze did not need anyone to organise it.

Two Days

Nickel closed at $29,770 a tonne on Friday 4 March. On Monday 7 March it rose 66 per cent and closed at $48,078 — a move that, on its own, would have been the largest in the contract's history. Margin calls went out that evening against positions marked at the new level, and Tsingshan and its brokers were now being asked for cash against a paper loss running into billions.

Covering a short means buying. Buying in a market with thin deliverable stocks and no natural seller means the price goes up, which increases the loss, which increases the margin call, which forces more buying. Each of the participants in that loop was behaving rationally.

LMEselect, the electronic platform, opens at one in the morning London time to serve Asian hours. Between the open and six o'clock on 8 March the three-month price went from around $50,000 to $101,365, a rise of more than 100 per cent in roughly five hours, in a session with a fraction of the liquidity of the London day.

The exchange operated volatility controls — static and dynamic price bands, designed principally to catch mistyped orders. They did not stop this. Oliver Wyman's independent review, commissioned by the LME and published on 10 January 2023, put the finding without decoration:

"The LME's price volatility controls did not control price volatility during the events."

The review also concluded that the existence of large, exposed short positions was in part due to the LME's inability to identify and address these positions as they were built up, and traced that blindness to the fragmentation of the position across multiple clearing members and between on-exchange and over-the-counter books.

Trading was suspended at 08:15 GMT. Notice TRADING/22/055, issued that day, set out the criteria for resumption and the price limits that would apply.

The Arithmetic Behind the Cancellation

What the exchange said afterwards about its reasoning is more interesting than the suspension itself, because it was not about fairness to traders. It was about whether LME Clear would survive the settlement cycle.

Had the 8 March prices been allowed to stand, LME Clear would have needed to collect $19.75 billion in margin from 28 banks and brokers — more than ten times the largest daily call in its history. Adrian Farnham, then chief executive of the clearing house, later described what he believed would have followed. At least seven clearing members would have defaulted. LME Clear would have taken a loss of around $2.6 billion, exhausting the defaulters' contributions and requiring at least $1.22 billion in further contributions from surviving members, which would probably have pushed another five into default. Farnham's term for the sequence was a "death spiral", at the end of which, on the exchange's own account, the LME would no longer have functioned as a venue for non-ferrous metals at all.

Some of the exposure came from how the LME is built. Alone among major derivatives venues it trades daily prompt dates rather than a handful of standardised expiries, a structure inherited from the sailing schedules of the 1870s, when a cargo of Chilean copper took three months to reach London. That design serves industrial hedgers who need metal on a specific day, and it scatters open interest across hundreds of contract dates in a way that makes any single participant's aggregate position harder to read. Add positions held bilaterally off exchange, invisible to the LME before 2022, and a short of 100,000 tonnes or more could sit in the market without appearing anywhere as a single number.

This is the structural point the episode is really about. A clearing house exists to mutualise counterparty risk, and it does that by converting market moves into cash demands on its members within hours. Craig Pirrong's work on the economics of central clearing had made the argument well before 2022 that clearing does not eliminate risk but transforms it into liquidity risk, concentrated at the moments when liquidity is hardest to find (Pirrong, 2011). Jon Gregory's study of central counterparties reached the same conclusion from the mechanics: a margin model calibrated on historical volatility becomes a procyclical amplifier in precisely the tail it was built for (Gregory, 2014). Duffie and Zhu's earlier analysis of whether a central counterparty actually reduces counterparty risk had shown that the answer depends on how positions are netted across products and members, which is another way of saying it depends on exactly the visibility the LME did not have (Duffie and Zhu, 2011).

The LME's answer was to refuse the cash demand by deleting the prices that generated it. That kept the clearing house solvent and transferred the loss to whoever had been long and right that morning.

Reopening, and the Bill

Notice TRADING/22/064 on 14 March announced resumption. That same day Tsingshan reached a standstill agreement with a consortium of banks led by JPMorgan, under which its lenders agreed not to call margin or close out positions while a secured liquidity facility was arranged. The paper loss at the peak had been around $8 billion. By the time Xiang unwound the position over the following months, the realised cost was estimated at roughly $1 billion, and Tsingshan remained intact and expanding.

The market reopened on 16 March and immediately hit a 5 per cent limit down, last trading at $45,590. On 17 March, with the limit widened to 8 per cent, it fell to $41,945 and stopped again. On 18 March it hit a 12 per cent limit. Matthew Chamberlain, the LME's chief executive, said the exchange had "deliberately prioritized stability" in setting narrow limits and would widen them once it observed what he called a "more orderly market". Traders who had spent a week unable to close positions at any price took a different view of which quality had been prioritised.

Litigation followed. Elliott Associates claimed $456 million and Jane Street Global Trading $15.3 million, both by judicial review, arguing the cancellation was unlawful and that it had expropriated their property. On 29 November 2023 the Divisional Court found for the LME and LME Clear on every ground, holding the decision lawful and rational and within the exchange's rules. The court did draw a distinction the claimants had fought over: Jane Street's trades amounted to possessions under Article 1 of the First Protocol to the European Convention on Human Rights, while Elliott's, being contingent agreements to trade rather than client contracts, did not. Elliott appealed, and the Court of Appeal dismissed the appeal on 7 October 2024.

Regulators moved more slowly and landed harder. The Bank of England, which supervises LME Clear as a recognised central counterparty, announced supervisory action in March 2023. The Financial Conduct Authority, which regulates the exchange itself, issued a Final Notice on 19 March 2025 imposing a penalty of £9,245,900 for failure to maintain orderly trading — the first enforcement action the FCA had ever taken against a Recognised Investment Exchange. The LME settled early and took a 30 per cent discount on the figure.

What the Episode Left Behind

Oliver Wyman's 27 recommendations built on measures the exchange had already taken under pressure: 15 per cent daily price limits across metals, and over-the-counter position reporting for all physically delivered contracts, which closed the specific gap that had let a position of Tsingshan's size assemble unseen.

Corners and squeezes are the oldest failure mode in commodity markets, and the LME had lived through them before. The Hunt brothers' attempt to buy the world's silver in 1980 ended with an exchange changing its rules mid-game to force liquidation. Sumitomo's copper trader hid ten years of losses on the LME before the exchange and its regulators noticed. What was different in 2022 is that the metal was almost incidental. The binding constraint was the cash the clearing house would have had to raise by lunchtime, in the same way that the gilt market's forced selling in September 2022 was about collateral calls rather than any revised view of British sovereign credit, and that Herstatt's failure in 1974 was about the hours between one leg of a settlement and the other.

A hedge that could not be delivered into produced a price that could not be paid for, on an exchange whose risk systems could not see the position because it sat in pieces across a dozen books. Xiang Guangda kept his company. Elliott lost in two courts. And the London Metal Exchange, having decided that eight hours of its own price history were a liability rather than a record, paid £9.2 million for the privilege of erasing them.