Amaranth Advisors and the natural-gas spread bet

A fund built on one trader's book

Amaranth Advisors was a multi-strategy hedge fund founded in September 2000 by Nicholas Maounis, based in Greenwich, Connecticut. By the mid-2000s its energy desk, run out of Calgary by trader Brian Hunter, had grown into one of the largest natural gas books in the market. At the end of December 2005, Amaranth's net assets stood at over $6 billion, with natural gas futures positions alone valued at over $5 billion. By some accounts the fund's assets under management peaked at around $9.2 billion in 2006, shortly before it collapsed.

A bet on the shape of the curve, not its level

Hunter's trade was not a simple wager that gas prices would rise or fall. He ran a calendar spread: long winter delivery months, chiefly March, and short the adjacent non-winter month, chiefly April, repeated across several contract years out to 2010. The premise was that cold-weather demand and tight storage would make winter gas more expensive relative to the following month, widening the spread between the two. Through 2005 and into 2006, this worked. The trade made Amaranth's energy book, and Hunter personally, very large sums: his 2005 compensation alone was reported at over $100 million.

A position too large for its own market

By July 2006 the position had grown to a scale that US Senate investigators later highlighted directly. Amaranth's combined NYMEX and ICE positions for January 2007 delivery reportedly exceeded 80,000 contracts, a volume that US Senate investigators said was roughly equal to the entire natural gas volume that American households used in that month. This scale meant Amaranth was not simply a large participant in the natural gas futures market. In some winter contract months, it was reported to hold a large share of all open interest, meaning a meaningful share of every contract outstanding.

The spread moves the wrong way

Through September 2006 the gap between winter and summer gas prices compressed sharply rather than widening. By 18 September, Amaranth told investors its natural gas losses for the month had reached around $3 billion, roughly a third to a half of the fund's assets depending on the AUM base used. The fund tried to hold the position, hoping the market would turn back in its favour, but daily variation margin calls on its exchange-cleared futures kept demanding cash the fund did not have to spare while the spread kept moving against it. On 19 and 20 September 2006, Amaranth transferred its entire energy derivatives portfolio to JPMorgan Chase and Citadel, reportedly at a discount to its own mark-to-market value that day. By 29 September, founder Nicholas Maounis confirmed to investors that the fund was winding down. Total natural gas losses for the year were reported at about US$6.6 billion.

A second reckoning, in the courts

A second strand of this story played out in the courts rather than the market. In July 2007 the US Commodity Futures Trading Commission (CFTC) filed a civil complaint alleging that on two NYMEX expiry days in 2006, 24 February and 26 April, Hunter had instructed colleagues to build up large futures positions ahead of the exchange's closing range and then trade heavily into the close, in an attempt to move the settlement price in Amaranth's favour. Internal instant messages quoted in the complaint include an instruction to "make sure we have lots of futures to sell MoC [market on close] tomorrow". The Federal Energy Regulatory Commission (FERC) added a third date, 29 March 2006, in a related action. These cases were resolved years later, and by then Amaranth's capital was long gone. In August 2009, the Amaranth entities settled with FERC and the CFTC for a combined $7.5 million, a fraction of the roughly $291 million FERC had originally sought, and the settlement order explicitly noted that the reduction reflected the fund's diminished assets since the 2006 collapse. Hunter himself settled personally with the CFTC in 2014, agreeing to pay $750,000 and accepting a permanent bar from trading during exchange closing periods.

Amaranth's position is best read as a bet on the shape of the futures curve, not on the level of gas prices. Holding a calendar spread, long one delivery month and short another, means the outright price of gas can rise or fall without much affecting the position, so long as the two legs move together. What matters is the gap between them. That makes it a pure basis trade in the Lecture 3 sense: the risk being traded is the relationship between two related prices, not the price of the underlying commodity itself.

This kind of spread is normally lower risk than an outright futures position, because both legs move with the same underlying commodity and much of the common price risk cancels out. But the spread itself trades in its own market, with its own supply and demand driven by storage capacity, weather forecasts and the positioning of other traders. When one participant's holdings become a large share of that market, the "safer" spread position becomes dangerous for a different reason: its own size. Amaranth's winter-month holdings were reportedly a large fraction of open interest in some contracts, which meant that exiting would itself move the spread further against the fund.

Two further mechanics from the course apply directly here. Margin is a demand for cash, not just a credit safeguard. NYMEX and ICE futures are marked to market daily, and an adverse move must be paid in cash within days, not absorbed as a paper loss over time. As the spread compressed through September 2006, Amaranth faced variation margin calls it could not keep funding while still holding a position too large to unwind quietly. A position can also become too large to exit even if the underlying view behind it is not obviously wrong. Amaranth's problem was not that natural gas futures were a flawed instrument. NYMEX contracts are standardised, exchange-traded and centrally margined, exactly the kind of contract this course treats as reliable market plumbing. The failure was a position built at a scale the market for that specific spread could not absorb on the way out.