Gresham Alternative Commodities

Finding stronger trends and weaker correlations

Hamlin Lovell
Originally published on 30 September 2021

The Gresham Alternative Commodities Absolute Return (ACAR) strategy has received The Hedge Fund Journal’s 2021 CTA Award for Best Performing Strategy over 3 years ending in December 2020, in the Trend Follower – Alternative Markets category. The awards were powered by performance data from Preqin.

ACAR’s most profitable trades in 2020 included both long and short exposures. Most of the returns came from short exposures in the first quarter, and long exposures in the fourth quarter, while the second quarter saw negative performance as the models moved from short to long. Key contributors in the second half of 2020 included iron ore and coking or metallurgical coal, driven by Chinese demand for steel; selected Chinese commodities, and various regional US electricity contracts. Performance has accelerated in the first half of 2021, as commodities such as European and Asian natural gas, and thermal coal, have reached record levels and catapulted through bank and broker price targets.

Gresham seeks markets that are driven by fundamental supply and demand and in particular by inelastic supply and demand, which can contribute to more extreme trends.

Gresham is targeting a long run Sharpe ratio in the region of one, which is much higher than many managers of liquid, directional trend following strategies who aim for a Sharpe between 0.2 and 0.5. Indices of such managers, such as the SG Trend Index, have in recent years delivered a standalone Sharpe of around 0.2 but could enhance a portfolio Sharpe. ACAR could also provide a diversification benefit versus more traditional CTAs, hedge funds in general, and long only indices of commodities as well as conventional asset classes such as bonds and equities. 

Finding the right markets 

The higher Sharpe target partly reflects a different perspective on trend following: “The historically typical view was that momentum produces a weak signal and a low Sharpe at the individual market level, so that the investment universe should be as large as possible to increase diversification and maximise the chances of identifying some trending markets. This was partly based on the Grinold and Kahn fundamental law of active management which stated that information ratios are proportional to skill multiplied by breadth,” says Scott Kerson, Head of Systematic Strategies at GreshamQuant, who was formerly Head of Commodities at Man AHL. 

The new paradigm that Gresham espouse is that they have stronger conviction in identifying markets that have both a higher propensity to demonstrate persistent trends, and lower and more stable patterns of correlation: “We believe we can do better than theoretical maximum diversification by using economic logic and statistics to reduce the investment universe to a higher standalone alpha per market and a more stable correlation, based on simulations and statistics. Our selection of markets is rooted in structural economic logic based on a micro-economic level understanding of each market,” says Kerson. 

Gresham groups ACAR’s current universe of markets into six categories: farm products; industrial materials; bulk including freight, coal and iron ore; European power; US power; and oil and gas. Gresham seeks markets that are driven by fundamental supply and demand and in particular by inelastic supply and demand, which can contribute to more extreme trends. Gresham can usually identify industrial and physical producers and consumer users (as opposed to investors and speculators) as the key market participants, which means that real world physical production and consumption constraints are most important. Indeed, the fact that some commodity markets are local rather than global is another source of diversification. “Power is not storable or fungible. Electricity markets can be locally bounded by interconnection networks. There are no wires over the Atlantic or over the Colorado Rocky Mountains and Texas has a rather unique local state power market. Natural gas can in some cases be local at particular hubs, though LNG is a global market subject to freight costs,” points out Kerson.

108

ACAR’s investment universe grew from 43 markets at launch in March 2017 to 108 markets by June 2021

Correlations between the markets traded are reduced by multiple dimensions of diversification. Gresham trades commodities in 15 countries and 5 continents: Africa, Asia, Europe, North America, and Australia. They trade over different contract periodicities: winter versus summer, monthly and quarterly contracts and some markets even stretch out for several years. They trade over time frames from days to weeks to months, with average portfolio turnover of 12 times a year. 

The ACAR markets have an average pairwise correlation of 0.2 over a 10-to-20-year history, and importantly this has been relatively stable, even during some stressed situations. “There was some increase in correlation in March 2020, but it was much smaller than that seen in financial market futures,” says Kerson. Most of these markets are inherently idiosyncratic, even if some of them may share common big picture drivers such as China being the largest consumer. Currency risk for some contracts not denominated in US Dollars is broadly hedged and it is very unlikely that any residual currency risk would lead to a meaningful currency carry trade. 

ACAR generates additional diversification by trading both long and short. Though the signals have been virtually all long throughout 2021, regardless of location or maturity, over time they should be short as often as long and average at zero directional bias.

Negative prices and hundred-baggers?

The most spectacular volatility in commodity markets generates a lot of headlines but is rarely seen in the futures markets that ACAR mainly trades. Commodity prices do occasionally go negative, though this is more common in spot than futures markets. “Though we have not yet seen a negative print in any futures contract we trade, after WTI oil went negative in April 2020 we did future proof models with safety belts and failovers to handle this,” says Kerson. Some commodity prices can also jump by as much as 10,000 percent in a matter of days: Texas ERCOT electricity prices multiplied one hundred-fold in February 2021, but again this happened very briefly in the spot market, and not in the futures markets that Gresham trade. 

Access and execution edge

ACAR’s markets are harder to access, and potentially more expensive to trade in terms of transactions costs and market impact at the strategy level, and operational resources at the firm level. This barrier to entry is rewarded: “We earn a complexity and implementation premium. It takes time and effort to collect historical and current prices, and when trading OTC there is also tactical game theoretic element in maintaining good prices and liquidity,” says Kerson. 

The markets traded include futures, OTC swaps to access futures e.g. in China, and “pure” OTC markets with no futures. “Though very little is pure OTC, the vast majority require high touch execution for efficient trading, and at least half cannot be traded electronically. Even where they can be traded electronically, you would not want to,” points out Kerson.

An element of counterparty risk is needed to access some of these markets. Virtually all OTC markets outside China are cleared, to manage the credit risk of books, but China is a special case where clearing of the onshore futures would require trading onshore in China. “This is not currently feasible because not all Chinese commodity futures have become QFII eligible, due to regulatory sensitivities around foodstuffs, coal and regulations. Additionally, there are concerns about expropriation and repatriation risk. Therefore, Chinese commodity futures are accessed through swaps,” says Kerson.

The choice of markets, how to combine them and how to budget risk is more important than the signal generation.

 Scott Kerson, Head of Systematic Strategies, GreshamQuant

Modern trend following 

“The choice of markets, how to combine them and how to budget risk is more important than the signal generation,” argues Kerson. Though identifying the right markets and accessing them at reasonable cost are the two core alpha drivers, Gresham’s models also have some distinguishing features that combine traditional and modern elements of trend following. 

Though fundamental analysis feeds into the selection of markets to trade, the trend models used to trade them are technical: based purely on price, and price derivative data such as volumes and volatility. The methodology is univariate trend following, so signals for each market are derived only from their own price history; there are no cross markets signals. The approach is also quite typical in that the portfolio aims to keep volatility within a range (of around 6 to 15%) and individual positions are sized inversely to volatility. 

As such, ACAR is relatively traditional in being a purely systematic technical trend follower, but it uses modern technology and techniques. “Over the decades, the technical details of how trend models are implemented have changed. For instance, rather than simply trading a binary long or short position of the same size, the strength of signals determines conviction that maps onto risk budgets and position sizes and aims to fine tune entry and exit. The model might use trend exhaustion indicators to reduce a position size and take some profits before a trend reverses but will not trade against the trend. Smoothing of entry and exit is also important for minimizing transaction costs. These features derive from the concept of confidence intervals around a forecast,” says Kerson. At any point in time, there may be no active position in 5 to 10 of the 108 markets traded.

Seasonality and term structure are important for commodity markets and many commodity strategies are dimensions of diversification for ACAR – but are not explicit inputs for signal generation. “Seasonality is considered in that differences between winter and summer natural gas informs the models but is not a signal per se. Term structure can be an important driver of returns but is not an explicit signal input. ACAR might sometimes generate either positive or negative carry via the term structure – so the carry element of returns might either augment or offset the price action,” explains Kerson. Separately from signals, determining roll dates as part of execution is a non-discretionary process that considers factors including term structure. 

Portfolio construction has also moved beyond mean/variance optimization and covariance matrices to allow for more uncertainty around how ephemeral alpha is, changing patterns of correlation, and the trade-offs between alpha and transaction costs. None of this currently incorporates machine learning but it may do in future. “If any machine learning is added it is most likely to be in portfolio construction rather than signal generation,” says Kerson. 

ACAR might at some stage go beyond purely directional trading. Currently, the strategy could accidentally end up with product, geographic or calendar spread trades across two or more of the markets it trades independently. These relative value trades are not currently part of the strategy repertoire, but they are being researched as potential additions. This research stream is also looking at hybrid and synthetic markets that could combine, or trade differences between, more than one market. 

Adding new markets

Commodity markets are the central focus of R&D for now. “The highest priority area of research is identifying new markets, and the second highest is managing and minimizing the frictional transaction costs of implementation,” says Kerson. 

The investment universe has already grown from 43 markets at launch in March 2017 to 108 markets by June 2021. ACAR has also ceased trading a small number of markets for reasons such as deteriorating liquidity. 

Weightings amongst the six sectors do not gyrate greatly but are reviewed when new markets are added. “The last material shifts were a modest addition of onshore China markets in mid-2020 while a larger change was made in 2021 adding hydrocarbons related to petroleum derivatives, such as Asian motor gasoline and Northwest Europe fuel oil. These increase diversification and potential capacity without detracting from alpha diversification,” says Kerson. 

New markets might at some stage free up capacity in the ACAR strategy, which is not constrained by Gresham’s long only commodities business since there is virtually no overlap between the markets traded.

New markets need to have economic properties that could lead to a higher Sharpe and low correlation to the existing portfolio, but they also need reasonable liquidity. Finding the right balance is paramount. The strategy is to some degree earning an illiquidity premium as an economic reward for trading and providing liquidity in more liquidity-constrained markets, but their liquidity needs to be adequate.

New markets also need to be capable of generating returns without much leverage: “Different grades of crude oil can show volatile differentials but would need a lot of leverage to make good profits from,” says Kerson. 

Gresham has started trading European and California carbon emissions (listed on ICE) and is researching other carbon contracts as well as even newer futures markets. Some cryptocurrencies could have supply and demand dynamics that naturally fit into the alternative commodities space: “The marginal cost of bitcoin is based on electricity prices, which can in turn be linked to other commodity prices. There are highly inelastic supply and demand curves and also a term structure of backwardation and contango. Prices need to do the work to clear the market,” says Kerson. 

Equities or ETFs could be used for the idiosyncratic risk of commodities that could not otherwise be accessed. “Lithium and cobalt are practically impossible to trade either OTC or on exchange, so one alternative could be dedicated listed companies with high betas to those metals. Rare earths, uranium, concrete, bulk paper, salmon fisheries, could be other examples,” says Kerson.

ACAR might use financial markets to gain commodity exposure but is not seeking esoteric or lowly correlated financial market exposures. Some alternative markets CTAs include financial contracts such as corporate credit or emerging market interest rates, but Gresham contemplates trading these in a separate strategy. “These are really interesting contracts and appropriate for alternative markets CTAs. But we have a very clear well defined investment thesis that hinges critically on the unique characteristics of markets with physical producers and inelastic supply and demand, which are unique to commodities. Also, some credit and emerging FX can introduce equity-like risk,” says Kerson.