Hedge funds are a little like referees and umpires, you don’t pay attention unless they do something wrong. Well, since the collapse of Long-Term Capital Management, no hedge fund has received as much ink as Amaranth Advisors. Losing $6 billion dollars in a little more than a week’s time will do that.
Much of the early attention focused on boiler plate warnings regarding hedge fund risk coming from usual sources. While something seriously went wrong with this $9 billion fund, its high profile investors were too smart to have invested in a fund that would lose 65% of its value in a couple of weeks on a simple long natural gas bet. That is not to say that there weren’t warning flags that were missed.
For example, Amaranth was purported to be a multi-strategy hedge fund. Multi-strategy funds don’t lose 10% in a month in one market, let alone 65% in a week. Amaranth appeared to have fallen in love with the energy sector and its all-star 32-year-old trader Brian Hunter. Another warning should have been its reported double-digit drawdown in May.
Hilary Till, principal at Premia Capital Management LLC and research associate at the Edhec Risk and Asset Management Research Centre, attempted to deconstruct Amaranth’s positions in a paper for Edhec. Till compiled all of the public information on Amaranth and was able to come up with a returns-based analysis of the fund’s energy strategy.
Simply stated, Amaranth had a natural gas spread strategy that stood to benefit from a couple of different weather related scenarios. They shorted summer and fall gas contracts and went long winter contracts. The strategy would have benefited from another active hurricane season (see “Long hurricanes,” above). Till suspected they were also long March gas and short April gas, a position that would benefit from an unseasonably cold winter as storage facilities would have become depleted at the tail end of the season (see “Chasing spring,” below). Amaranth appears to have put these positions on for every year out to 2011.
Till says there were two rationales for the fund’s strategies. “The rationale for the bear-calendar spreads is related to the tendency over the last two years for the calendar spreads to go beyond full-carry levels by September. About a week after Hurricane Katrina hit the bear-calendar spreads became profitable because of the disruption to seasonal storage operations. The curve eventually went into further contango. The rationale for the March-April spreads may have been as follows. The March contract expires at the end of February. If one has an extremely cold winter, even if one starts out with a lot of natural-gas-in-storage, it is possible to have a stock-out situation by the end of February. If this scenario occurs, there is no cap on how much March can rally with respect to the rest of the curve. The March-April spread may be interpreted to have historically had a long-option-like payoff profile.”
She notes that while there may have been solid fundamental reasons for both positions, they are not a natural hedge. In fact it could be argued given what happened, that the two basic positions were somewhat correlated. With no hurricane related supply disruption, the chance that there would be an end of winter supply shortage would be lessened. Front month natural gas prices had dropped more than 66% from December 2005 to July 2006, so supplies were building. In hindsight you could see people were banking on another severe hurricane season. The 2006 summer/winter spreads went from a contango of under $1 at the beginning of the year to more than $4 by late summer (see “Something strange,” below). At that point it is fair to conclude that a bad hurricane season was priced into the market and the possibility that there would not be a hurricane-related supply disruption presented a greater risk of a volatility spike to the market. In effect, Amaranth bet on a hurricane with a series of summer/winter spreads out several years, and bet on an unseasonably cold winter with a series of March/April spreads also out several years.
Erk Hinrichsen, senior managing director of Energy Arbitrage Management, says, “It looks like they put the same position on over and over. Yes, you could call it arbitrage — long natural gas up front and short the deferred, and long natural gas in the wintertime and short in the summer time. And he did the same trade over and over again at huge sizes.”
One thing seems clear, regardless of whether the trades were smart or dumb or justifiable, the position sizes were not. In the conclusion of her analysis Till noted, “These strategies were and are economically defensible, but the scale of their position sizing relative to their capital base clearly was not.”
Till wrote that a value at risk/recent volatility analysis as of the end of August would have underestimated Amaranth’s risk during a liquidation pressure event. This would particularly be true in the spreads going out four and five years.
“Where were the risk controls?” Hinrichsen asks. “Isn’t there a limit on the size of the position that you could put on? We are very strict; one position can only be 5% of the total equity of the portfolio. It would have prevented it, a very simple thing.”
John Alden, principal of energy based Spinnerhawk Natural Resources, says that volatility could have thrown off certain value-at-risk measures. “They probably had a look back of a year or two,” Alden says.
In fact the greater amount of data the fund used in their risk analysis may have masked the true risk because the spreads had moved so deep into cantango in 2006.
Michael Boren, principal of McComber Energy Fund, noticed increased volatility in energy markets and energy based equities. “You can be right about the weather and wrong about the market,” Boren says.
Hinrichsen agrees, “Your position sizes should be smaller because the volatility is bigger. It used to be that if a crack spread (buying heating oil against crude), if a spread moved 25¢ to 50¢, it was a big move, but now we have that almost daily. So if you have that much more volatility, wouldn’t it be the prudent thing to say ‘fine I should reduce my size because the dollar impact is now the same.’ If beforehand I am looking to make $1 on a crack and it takes time, nowadays it moves that much in a day.”
Amaranth wasn’t trading crack spreads, but the volatility of what they were trading had changed dramatically throughout the first seven months of 2006, and if they ran stress testing on the historical volatility of those spreads, it would not have captured all of the risk. They were in a new environment in terms of natural gas calendar spreads, and appeared to have stress tested those positions against historical moves that did not reflect the current volatility.
When energy based hedge fund Mother Rock LP had to wind down a month earlier, a source close to the fund pointed out that certain spread relationships had seen volatility well in excess of anything the market had seen in the past decade. That would mean, risk measures based on those past 10 years would not reveal the true risk to those positions.