Amaranth Advisors

Analysis and Commentary

The crisis at Amaranth Advisors was caused by a single package of trades that went wrong. Using what is known as a “calendar spread strategy,” head energy trader Brian Hunter bet that the spreads between winter-month natural gas contracts and non-winter-month contracts would widen. Instead, they shrank. Using futures and options, Hunter bet more than half the firm’s capital on this single thematic idea. And because the firm bet so many of its assets on this one trade, it apparently prepared for only one scenario: we’re right and the market is wrong. Whatever the exact rationale, the strategy was highly aggressive and it failed.

There are (at least) two ways to look at this tragic trade — (1) as a single investment decision that would either succeed or fail on its own merits or (2) as one component in a broad trading strategy based on exploiting market anomalies. As a single investment decision, Hunter’s strategy failed to have a sound basis. Even though a back-tested version of the calendar spread strategy that Hunter used had worked in 14 of the previous 15 years, with only modest losses in the single losing year, he had no basis to believe that this strategy would work this particular time. In short, the only valid claim Hunter could make is that he observed a persistent pattern in gas prices. He made no claim as to why this this pattern existed, nor any claim to the underlying cause and effect.

Although the natural gas market does exhibit certain predictable patterns over time —  namely, that summer demand is light and gas prices tend to fall, while winter demand is heavy and prices tend to rise — these patterns can easily be disrupted by shocks to the system, changes in weather patterns, economic growth, recessions, and so on. Hunter’s investment was based on the assumption that the historical pattern in natural gas prices would continue. It was simply a correlation trade. In fact, in September of 2006, the natural gas market did something it hadn’t done before. As natural gas inventories rose and the threat of another harsh hurricane season receded (2005 was a particularly harsh year for hurricanes), natural gas prices dropped. And winter-month contracts dropped much faster than summer months. Hunter’s strategy was wrong.

One of the overarching lessons here is that correlation is not causation.  As such, trades based on observed correlations must be diversified enough for the law of averages to work.  This means that correlation trades are an acceptable form of investing only insofar as the investor concedes that he or she knows little (or nothing) about why the patterns might persist, revert to the mean, etc.  Consequently, the investor in correlation trades must use small positions sizes and diversify risk across a large number of risk exposures. In contrast, trades based on understanding of cause and effect or an attractive upside-to-downside relationship can be undertaken with greater confidence. Though such trades still must be diversified to reflect the possibility of error in facts or judgement. In all cases, the investor must ensure that positions fit within the portfolio’s stated mandate, the viability of the firm is not jeopardized by the failure of a given trade, and the standards of prudence and fiduciary duty are met. In short, Amaranth “bet the farm” on correlation trades and hence failed on all these counts.

In making the bet, Hunter fell into the trap of status quo bias — that is, expecting past market behavior to persist in the future. Perhaps the large gains that Amaranth achieved in natural gas contracts the previous year led to a blinding greed trying to expand that success. Hunter failed to appreciate the difference between correlation and causation with respect to natural gas prices. Having found a historical anomaly, Hunter didn’t have an actionable item as a singular investment without understanding why the phenomenon might continue. Only by establishing causation could Amaranth make an actionable investment decision with a sound basis.

Alternatively, if these investments are viewed as part of a broad trading strategy to exploit anomalies (such as price patterns in various natural gas contracts), the strategy still failed, but for different reasons. In the case of trading on anomalies, traders must openly accept that they do not know why an anomaly exists, only that it does. In the absence of such knowledge, the trader/investor can only speculate as to whether or not the anomaly will continue. Therefore, a prudent investor/trader diversifies trades over a large pool of anomalies (ideally, among asset classes, geographical areas, and time). Speculating on such anomalies is an acceptable form of investment, but one’s position sizes must reflect ignorance about the reasons for the anomalies and therefore should remain a small percentage of the portfolio. Investors can find safety of principal only by exploiting large numbers of anomalies so as to diversify the risk. In this case, Hunter and Amaranth fund managers bet everything on a single theme. They failed to diversify appropriately and concentrated far too much risk in a single theme. With more than 50% of the portfolio invested in this particular trade, it was clearly overstated.

Although the calendar spread that Hunter used could theoretically support positions based on historical probabilities around natural gas prices, such a strategy is more likely to be successful if it meets certain conditions. A manager must acknowledge that he or she has imperfect insight into the future and must adequately assess whether the historical probabilities and payoffs are favorable for a particular strategy. The position should fit within the context of a given portfolio’s mandate, and the trade must be treated roughly the same as any other trade so that the law of averages has a chance to work. If it had worked, Hunter might now be famous for the amount of money he made on this call. Having bet so heavily, however, he virtually ensured a binary outcome without any middle ground. A sound trading strategy encompasses positions that probabilistically achieve both success and failure. In other words, an investor/trader must move away from focusing on a single expected outcome and begin thinking through possible outcomes or scenarios.

To compound matters, not only did Amaranth have a poor basis for establishing the positions that it did, the firm also ensured that it could not exit the positions in a timely way. By owning very large portions of specific natural gas contracts (in some categories, Amaranth owned more than 50% of a given contract), Amaranth in essence “became” the market for these contracts. What such a position meant in practice is that if the trades went wrong for any reason, Amaranth could not unwind its positions. Liquidity is required to exit any position, but it is acutely important during periods of stress — when not only do prices of securities perform poorly but also trading liquidity of the security deteriorates sharply, thereby making the contracts impossible to trade. Considering that Amaranth owned more than 50% of these troubled contracts, once Hunter’s position went south, nobody was willing to take the other side of these trades because they feared incurring the same liquidity problems. So, liquidity dried up and Amaranth could not exit at any price. Recognizing Amaranth’s distress, other hedge funds took the opposite sides of these trades, exacerbating the situation. Liquidity must be considered when evaluating risks related to exit strategies in times of stress (or as a potential opportunity should the investor be able to live with periods of stress if liquidity is mispriced).

Amaranth Advisors could have avoided all this grief by performing basic scenario planning to examine what would happen following various changes in underlying supply and demand for natural gas under mild, normal and severe weather scenarios. At the very least, the firm should have carried out what-if scenario analysis for adverse changes in market prices.

Amaranth and Hunter might have had the intent to manipulate prices in the natural gas markets (as regulators subsequently claimed), but we have not seen direct evidence to support this claim. Perhaps this crisis could have been avoided for Amaranth by not hiring Hunter in the first place. Human resource departments play an important role in understanding the nature of the people they hire. Of course, hiring Hunter may have been simply part of the firm’s desired trading strategy.

CFA Institute members: This is eligible for 0.5 CE credit.