Famous Futures Blowups: Barings, Metallgesellschaft, Amaranth and MF Global

The four most instructive futures disasters are Barings (1995), where one trader’s hidden positions sank a 233-year-old bank; Metallgesellschaft (1993), where a hedging program ran out of cash; Amaranth (2006), where a hedge fund became too big for the natural gas market; and MF Global (2011), where a futures broker used customer money to cover its own needs. Each failed in a different way, and the official reports into each are unusually specific about why.

In short: Barings failed because of a total breakdown of controls: Nick Leeson ran both the trading and the settlement of his own trades, hid losses in an error account, and reached cumulative losses of £827 million before anyone in London understood. Metallgesellschaft’s US subsidiary hedged long-term fixed-price oil sales with short-term futures, and when oil prices fell the margin calls arrived years before the offsetting profits. Amaranth built natural gas positions so large that, according to the US Senate investigation, it at times held 40% of all outstanding NYMEX winter contracts, and it lost about $6 billion in September 2006 when it could not get out. MF Global used customer segregated funds to meet its own liquidity needs in its final week; customers were eventually repaid in full, but only after years of litigation.

Why study blowups at all

Each of these cases involved regulated exchanges, large and sophisticated institutions, and positions that someone, at some point, believed were under control. They show how futures risk actually materialises: not as a gradual loss, but as a sudden mismatch between positions and the cash needed to carry them. The daily mechanics behind every one of them are in how futures margin works.

1. Barings, 1995: one trader, no controls

What happened

Barings was a London merchant bank founded in 1762. In Singapore, its subsidiary Baring Futures (Singapore), or BFS, executed trades on SIMEX (the Singapore exchange) and on the Osaka exchange in Japan. Nick Leeson, its general manager, was supposed to be running a low-risk business: executing client orders and arbitraging small price differences between the same contract in Singapore and Osaka.

According to the Bank of England’s Board of Banking Supervision report, published July 18, 1995, and summarised by the Reserve Bank of Australia, Leeson was instead taking large unauthorised positions:

• Long Nikkei 225 stock index futures
• Short Japanese government bond futures
• Short options on the Nikkei, a bet that the market would stay calm (a “short volatility” position)

He booked the losses in an error account numbered 88888, opened in July 1992, which was left out of the reports sent to London.

The numbers

Date	Cumulative unrecognised losses
End of 1994	Over £200 million (of which £185 million arose in 1994)
February 27, 1995	£827 million
After closing all positions	£927 million in total

In January 1995 the Kobe earthquake hit Japan and the Nikkei fell. Leeson’s long futures and short options both lost heavily as the index dropped and volatility rose. Instead of cutting, he added to the positions. By late February, the bank’s funding of BFS’s margin requirements had reached about twice the Barings Group’s capital. Barings was placed into administration on February 26, 1995, and the Dutch bank ING later acquired its business for a nominal £1.

What the report found

The Board of Banking Supervision did not blame derivatives as such. It attributed the collapse to “a failure on the part of a number of individuals … to do their jobs properly” and a “virtual total failure of risk management systems and controls.” Leaving Leeson in charge of both the front office (trading) and the back office (settlement and records) was, in the report’s words, “a most serious failing.” Senior management in London also did not appear to appreciate the margin requirements, and kept sending money to Singapore to fund them while believing the business was profitable.

The lesson

Segregate duties, and question profits you cannot explain. A desk reporting large profits from a supposedly low-risk arbitrage, while needing ever more cash for margin, is a warning sign. The report’s lessons included that “clear segregation of duties is fundamental to any effective control system” and that “management teams have the duty to understand fully the businesses they manage.”

2. Metallgesellschaft, 1993: a hedge that ran out of cash

What happened

Metallgesellschaft AG (MG) was then Germany’s fourteenth-largest industrial company. Its US oil subsidiary, MG Refining and Marketing (MGRM), sold long-term contracts to customers promising to deliver heating oil and gasoline at fixed prices for up to ten years. According to a 1995 Federal Reserve Bank of Richmond study by Anatoli Kuprianov, by September 1993 MGRM had committed to sell forward “the equivalent of over 150 million barrels of oil,” with most contracts running for ten years.

Those sales left MGRM exposed to rising oil prices. To hedge, it bought energy futures and swaps. Because liquid futures only extend a short distance into the future, it used a stack-and-roll strategy: hold a large stack of short-dated long futures and roll them forward each month. The general mechanics of rolling are in rolling futures contracts.

In late 1993, oil prices fell. MGRM’s long futures lost money, and those losses had to be paid in cash every day as variation margin. The offsetting gains on its fixed-price customer contracts were real on paper, but they would be collected over the following decade. At the same time, the oil curve moved into contango, so each monthly roll also cost money (see contango and backwardation).

The numbers

Kuprianov reports the losses were “later estimated at over $1 billion.” MG’s board dismissed senior management, liquidated much of the hedge and the customer contracts, and negotiated a $1.9 billion rescue package with its creditor banks.

Worked example: the cash-flow mismatch

Using round numbers close to the MGRM case. The real NYMEX crude contract is 1,000 barrels.

• Long futures hedge: 150 million barrels = 150,000 contracts (illustrative one-for-one hedge)
• Oil falls $1 per barrel
• Cash paid out in variation margin: $1 x 150,000,000 = $150 million, within days
• Gain on fixed-price sales: also about $150 million in present value, but received gradually over up to ten years as deliveries occur

A fall of $5 per barrel means $750 million of cash out, against gains that arrive slowly. The hedge “works” economically and can still bankrupt you.

The debate

Academics still argue about MG. Culp and Miller (1995) argued the strategy was sound and that the board’s decision to liquidate locked in the losses. Edwards and Canter, in the Journal of Futures Markets (1995), and Mello and Parsons argued the hedge was far too large; Kuprianov notes Edwards and Canter’s estimate that buying futures for about 61 million barrels would have minimised the variance of cash flows on a 160-million-barrel long-term exposure.

The lesson

A hedge must be fundable. Matching the economic value of exposures is not enough; you also need enough cash or credit to meet margin calls for as long as the mismatch lasts. Hedge size should account for the timing of cash flows, not just their total.

3. Amaranth, 2006: too big for the market

What happened

Amaranth Advisors was a multi-strategy hedge fund. Its energy desk, led by trader Brian Hunter, built enormous positions in natural gas futures and swaps, mainly long winter-month contracts and short summer or autumn months, a bet that winter-summer price spreads would widen.

The US Senate Permanent Subcommittee on Investigations examined millions of trades and published its staff report, “Excessive Speculation in the Natural Gas Market,” on June 25, 2007. Its findings included:

• At times Amaranth controlled about 40% of all outstanding NYMEX natural gas contracts for the winter season (October 2006 to March 2007), and as much as 75% of the outstanding contracts for November 2006.
• In May 2006, on 18 of 21 trading days, Amaranth held more than 50% of open interest in the January 2007 NYMEX contract, and on every trading day at least 55% of the November 2006 contract. In the report’s words, “Amaranth was too big for the market it had created.”
• When NYMEX told Amaranth to reduce its positions in the September and October 2006 contracts in August 2006, Amaranth moved the positions to the Intercontinental Exchange (ICE), then an unregulated venue for equivalent “look-alike” swaps, keeping its overall exposure.

The losses

The spread trade first went wrong in May 2006. The Senate report says that in the last week of May the January/November spread fell from $2.15 to $1.73, about 20%, and Amaranth lost more than $1.16 billion on its energy positions that week, after building its January/November spread position to nearly 60,000 contracts.

Worked example: what a spread move costs at that size

The real NYMEX Henry Hub natural gas contract is 10,000 MMBtu; a $0.001 tick is worth $10.

• Spread change: $2.15 - $1.73 = $0.42 per MMBtu
• Per spread (one long and one short contract): $0.42 x 10,000 = $4,200
• On 60,000 spreads: $4,200 x 60,000 = $252 million

That is the loss on one spread position from one week’s move, before counting Amaranth’s other positions. The fund kept trading, recovered for a time, and then in September 2006 the winter-summer spreads collapsed again. Amaranth lost about $6 billion in a little over a week, roughly two-thirds of its assets, and closed down. Hilary Till’s review of the Senate report describes the fund as having lost “about 65% of its $9.2 billion in assets.”

What the Senate report concluded

The Subcommittee found that Amaranth’s trading “significantly distorted” natural gas prices and recommended that position limits apply consistently to both regulated futures and equivalent swaps. That recommendation fed into later reforms; the current framework is described in position limits and large trader reporting.

The lesson

Liquidity is a function of your own size. A position that dominates a market can be marked at prices that no one will pay when you try to sell. Amaranth’s own investor letter called May 2006 “the worst month since inception,” and its traders found that buyers disappeared when they tried to take profits.

4. MF Global, 2011: when the broker uses customer money

What happened

MF Global was a large futures commission merchant (FCM) and broker-dealer, run from 2010 by Jon Corzine, a former head of Goldman Sachs and former governor of New Jersey. In 2011 the firm took large positions in European government bonds. As concerns about those positions grew in October 2011, the firm faced a liquidity crisis, and it filed for bankruptcy on October 31, 2011.

US rules require an FCM to keep customer funds segregated from its own money. The report of the SIPA trustee, James Giddens, found that segregated funds “were used to fund MF Global’s liquidity needs in at least the latter part of the week of October 24th,” and that the firm ultimately reported an approximately $952 million shortfall in segregated funds. In Senate testimony, Giddens later described an approximately $1.6 billion shortfall in segregated customer property overall, including funds held for customers trading on foreign exchanges.

What the CFTC did

Date	CFTC action
November 2013	Consent order against MF Global Inc.: $1.212 billion in restitution to customers and a $100 million penalty
December 2014	Consent order against parent MF Global Holdings: restitution (jointly with MF Global Inc.) and a $100 million penalty
January 2017	Consent order against Jon Corzine: $5 million penalty, a permanent bar from registering with the CFTC or working for an FCM, for his role in the “unlawful use of customer funds totaling nearly one billion dollars” and failure to supervise
January 2017	Former assistant treasurer Edith O’Brien: $500,000 penalty

The CFTC said restitution paid under the orders satisfied all customer claims. In November 2013 a bankruptcy judge approved a plan to repay former commodity customers in full, funded in part by settlements with counterparties.

The lesson

Your counterparty is your broker, not just the exchange. The clearing house protects its members, but customers depend on their FCM keeping their money separate. MF Global led to tighter segregation and reporting rules. The current protections, and why futures accounts are not covered by SIPC, are in how futures customer funds are protected.

What the four cases have in common

Case	Year	Loss	Main failure	Lesson
Barings	1995	£827 million to Feb 27; £927 million total	No segregation of duties; hidden account	Controls and oversight
Metallgesellschaft	1993	Over $1 billion	Cash-flow mismatch in an oversized hedge	Hedges must be fundable
Amaranth	2006	About $6 billion in September alone	Positions too large for the market	Size destroys liquidity
MF Global	2011	About $1.6 billion customer shortfall (later repaid)	Misuse of segregated customer funds	Broker risk is real

Three threads run through all four:

1. Margin turns paper losses into cash demands. Every case reached its crisis through funding, not through a slow decline in value.
2. Size changes the risk. Positions that are small relative to the market can be exited; positions that dominate it cannot.
3. Oversight failed before the market did. Each firm had warning signs that someone with authority could have acted on.

What can go wrong for ordinary traders

Most readers will never hold a Barings-sized position, but the same mechanics apply at small scale:

• A hedge or trade that looks sound can still fail if you cannot meet margin calls.
• Doubling a losing position to “get back to even” was Leeson’s pattern.
• Thinly traded contracts can be impossible to exit at the screen price.
• Your money is only as safe as the firm holding it and the rules it follows.

The short version

Barings collapsed in 1995 because one trader controlled both his trades and their records, hid losses that reached £827 million, and was funded by a head office that did not understand margin. Metallgesellschaft’s oil subsidiary lost over $1 billion in 1993 because a stack-and-roll hedge on 150 million barrels of long-term sales demanded cash immediately when prices fell. Amaranth lost about $6 billion in 2006 after building natural gas positions that the US Senate found were, at times, 40% of the NYMEX winter market. MF Global used customer money in its last days in 2011; customers were repaid in full, eventually. Each disaster turned on the same point: leverage and daily settlement convert mistakes into immediate cash demands.

Related: Futures risks explained honestly · How clearing houses work · Negative oil prices in April 2020