Background:
Natural gas is a highly volatile & inflammable fuel.
Natural gas futures are even more volatile to trade. It is usual for NG to move 10% in either direction in matter of minutes. You don’t get “widow maker” as a moniker for nothing.
However, the higher volatility always provides greater opportunities for making profits. Making it alluring for adventurous traders. Let’s examine the Amaranth disaster.
Fundamentally, this trade too was a spread bet like Aas trade we saw in chapter 2. Specifically, called as “calendar spread” i.e. instead of betting against 2 different instruments, one can use same underlying but delivered in different months.
In Natural Gas, calendar spread trading involves the simultaneous purchasing of a futures contract for one month—called the “long” month, typically March (because spring comes and winter ends) and the selling of the same number of futures contracts for a different month—called the “short” month, typically April, as summer season begins.
March is considered the last winter contract, as it is the last month when utilities are usually pulling gas out of storage – while April is the first month when utilities start injecting gas back into storage. A wider spread indicates a big need for supply, while a small gap between the two means utilities think there’s enough gas in storage.
The spread between March and April futures — essentially a bet on how tight supplies of the fuel will be at end of North American winter.
Simply put, “long winter, short summer”. The trade bets that Mar contract will be unexpectedly higher due to hurricanes (cutting future supply) or severe winter sucking out more inventories than anticipated.
Buying the spread means, long March and short April. This was price situation as of 8th Sep 2021 for trade of 2022.
In the winter of 2020, the widow maker reaped rich harvest.
This trade is typically entered in summer of previous year to take advantage of the price dislocations in the futures term structure.
We won’t go into about the technicalities of the term structure here. But briefly,
- Prices tend to increase as we go deeper in time.
- Current month prices tend to ‘leak’ or ‘infect’ future months’ prices. This entire trade is based on assumption that at delivery time, the infection will cure itself.
- Other peculiarity of natural gas is that prices are higher in winter months (due to heating demand) and lower in summer (due to less cooling demand, relatively).
The trade:
Opening positions in June / July of 2006:
- buying futures contracts for January 2007 while selling futures contracts for November 2006—the “January/November spread “representing a bet that January prices would be much higher than November prices; and
- buying futures contracts for March 2007 while selling futures contracts for April 2007—the “March/April spread”—representing a bet that March prices would be higher than April prices, when demand for natural gas for home heating diminishes significantly.
The hidden flaw in the trade:
- A key flaw was the position size relative the market. Amaranth’s positions were so big, that it distorted the market.
The senate report states that “In late July, Amaranth held a total of more than 80,000 NYMEX and ICE contracts for January 2007, representing a volume of natural gas that is equivalent to the entire amount of natural gas eventually used in that month by U.S. residential consumers nationwide.”
“Amaranth’s large-scale trading was a major driver behind the rise of the January/November price spread from $1.40 in mid-February to $2.20 in late April, an increase of more than 50%.” - Huge leverage they had employed. 5x was just too big for a market as volatile as NG.
Trigger Event:
In late August, the market moved sharply against Amaranth. The amount of natural gas in storage was very high, and there had been no major hurricanes to disrupt production. The winter/summer spread positions that Amaranth had invested in during the spring and summer began to fall.
The price of the NYMEX futures contract to deliver natural gas in October 2006 fell from a high of $8.45 per MMBtu in late July to just under $4.80 per MMBtu in September, the lowest level for that contract in two and one-half years. The difference in price between the NYMEX natural gas futures contract for March 2007 and for April 2007 – i.e., the spread – fell from a high of nearly $2.50 per MMBtu in July to less than 60 cents in September, a drop of 75%.
Amaranth’s margin requirements grew to over $2 billion, and eventually reached nearly $3 billion. As a result, Amaranth no longer had the capital to buy large positions in the face of falling prices.
End game:
On September 18th, 2006, market participants were made aware of a large hedge fund’s distress. Amaranth Advisors, LLC, issued a letter to investors, informing them that the fund had lost an estimated 50% of their asset’s month-to-date.
Loss:
By the end of September 2006, these losses amounted to $6.6-billion, making Amaranth’s collapse the largest hedge-fund debacle to have thus far occurred.
Lesson checklist:
| Was the trade idea flawed? | No. This is common arbitrage trade in many markets especially energy including heating oil. |
| Position size | Yes. Too big relative to market size. Amaranth held as many as 100,000 natural gas futures contracts at once, representing one trillion cubic feet of natural gas, or 5% of the natural gas used in the United States in a year. At times Amaranth controlled up to 40% of all the open interest on NYMEX for the winter months (October 2006 through March 2007). Madness. |
| Concentration | HIGH. Single pair arbitrage increases risk. |
| Was it a ‘black swan’? | No. |
| Leverage | Yes. 5x leverage is extraordinary for an energy hedge fund. It was the most highly leveraged fund in the business. |
| Was it a fraud? | No. Just levered spread bet going wrong as they often do. |
What can we learn from this?
Simple: Do not trade NG futures.
NG calendar spread is not an arbitrage trade – even though it looks like one. It is a one directional bet on weather patterns. In 2022, war on Ukraine & explosion in Texas LNG terminal caused further disruptions.
That is beyond our skill.
We show price action of the Mar 23-Apr 23 spread, the classic widow maker below:
References:
- https://www.hsgac.senate.gov/wp-content/uploads/imo/media/doc/REPORTExcessiveSpeculationintheNaturalGasMarket.pdf
- https://www.reuters.com/markets/asia/collapse-us-natgas-prices-cut-widow-maker-spread-20-month-low-2021-12-01/
- https://frontmonth.substack.com/p/amaranthology
Postscript: Amaranth owned positions well into the future! Hard to speculate on the mindset of a person willing to take such positions – and total failure of risk management that allowed him to do this.
Senate report: Amaranth also held large positions in other winter and summer months spanning the five year period from 2006-2010. In aggregate, Amaranth amassed an extraordinarily large share of the total open interest on NYMEX. During the spring and summer of 2006, Amaranth controlled between 25 and 48% of the outstanding contracts (open interest) in all NYMEX natural gas futures contracts for 2006; about 30% of the outstanding contracts (open interest) in all NYMEX natural gas futures contract for 2007; between 25 and 40% of the outstanding contracts (open interest) in all NYMEX natural gas futures contracts for 2008; between 20 and 40% of the outstanding contracts (open interest) in all NYMEX natural gas futures contracts for 2009; and about 60% of the outstanding contracts (open interest) in all NYMEX natural gas futures contracts for 2010.
