BlackBird Asset Management – Commodity Arbitrage

Blending quantitative rigour with discretionary judgment

Hamlin Lovell
Originally published in the June | July 2020 issue

Nearly all commodity markets (bar palladium, gold and wheat) lost value in the first quarter of 2020, while relative value strategies across multiple asset classes have been struggling in the Great Coronavirus Crash (GCC). Yet relative value commodity trader, BlackBird Asset Management, had its best ever month in March 2020, up 5.89% in USD. The longer-term track record is also impressive, annualizing at 13.2% with volatility of 7.6%. And this sort of true alpha can still command hedge fund fees of 5% and 20% on the 2.5x levered version of the strategy, which has annualized 33.50% net of fees. 

Coronavirus has been a boon not only because the general spike in volatility has dislocated markets and thrown up more relative value opportunities, but also because the focus on staples has contributed some trade ideas. One was a bullish spread (long the nearby expiry and short the back expiry) on wheat, which is used (mainly) for food, versus a bearish spread (short nearby/long back) on corn, which is also used heavily for ethanol fuels where demand has fallen. “Wheat is a key strategic commodity for many countries so they will buy it defensively. We suspect that some countries are adding to their strategic food reserves. Russia and Romania have also limited exports to try and keep local prices low,” says BlackBird founder Olivier Merle (whose surname means blackbird in French). Meats were also profitable to trade in March. These trades profited from a steepening of the curve, but the opportunity did not last long: it had normalized by early May. 

Wheat is a key strategic commodity for many countries so they will buy it defensively. We suspect that some countries are adding to their strategic food reserves.

Olivier Merle, Founder, Blackbird Asset Management

BlackBird’s return profile has no meaningful correlation either to the BarclayHedge CTA nor BarclayHedge Agricultural Traders indices, because manager Merle is doing something rather unique. Neither a technical trend follower nor a purely discretionary fundamental trader, he has carved out a niche that uses discretionary market savvy to select, and trade, signals generated from a sophisticated quantitative process that employs both technical and fundamental inputs. The strategy builds on Merle’s education in engineering, statistics and econometrics, and his years of price action trading experience between 2011 and March 2017, before he set up managed accounts and a US fund, catering for US investors and Merle himself, who is the second largest investor in the strategy.

Smaller commodity markets

BlackBird mainly trades around 25 agricultural commodity futures – including oilseeds; meats; milk; grains and softs – listed on seven global exchanges, currently in the US, Canada, Europe and Asia (CBOT; CME; ICE; Euronext; Kuala Lumpur, Winnipeg and SGX). Merle reckons that this investment universe provides a richer opportunity set than can be exploited by some larger funds, which might not be able to trade some of the smaller commodity markets – or at least not in meaningful size, because many exchange listed commodity derivative markets have limited capacity. “In financial markets, derivatives are many times larger than cash markets. For commodities the cash market is generally much larger than the derivatives market for most commodities (with exceptions such as CBOT’s very specific wheat contract),” explains Merle. “Options on commodities tend to be much smaller than futures on commodities,” he adds.

Discretionary filter for quantitative models

Merle does not trade cash i.e. physical commodities markets, but he monitors them closely for intelligence. Eyes and ears on the cash markets – through dialogue with a network of traders built up during Merle’s time managing a quantitative department at Agritel advising cash traders – provide important information flow. “I need to understand how and where corn grows, where the flows are, how grains are traded around the Black Sea, which is important for wheat and a strong regional player for corn, and the issues cash traders are considering,” says Merle. At Agritel, he authored a weekly newsletter on agricultural spreads called “Highlight of the Week”.

A fairly symmetrical win: loss ratio around 50:50 means that Merle needs to increase his hit rate to around 60-65% to consistently profit. Discretionary analysis filters out over half of the signals that are thrown up by BlackBird’s quantitative models, which means that Merle’s real money track record is more informative than any back test alone might be. The quant models could in theory be back-tested to around 2000 using a database of cash and future prices, but the full BlackBird process – including the discretionary input – can only be reviewed on a live trading basis, back to 2010, in three stages. First, Merle’s price action trading account with Interactive Brokers. Second, managed accounts run for the RCube Commodity Spread strategy since 2017. And third, a US fund structure launched in 2019.


BlackBird mainly trades around 25 agricultural commodity futures – including oilseeds; meats; milk; grains and softs – listed on seven global exchanges.

The quantitative models are based on multiple data and model inputs, including mean reversion; lead/lag relationships; roll differentials; rollover patterns; seasonality; contango and backwardation. The term structure models tend to contain the largest number of factors. Merle formulates hypotheses mathematically and prioritises the strongest relationships, while guarding against overfitting.

Models need to keep abreast of market evolution. “We consider the behaviour of other market participants, including long only and passive ETFs and indices, but it is not straightforward to assume they will all be rolling the front month on a predictable date; the US OIL ETF is not the only product that has moved further out the curve. We also look at commitment of traders’ reports to gauge CTA positioning. An important new category for systematic managers is alternative risk premia. We estimate that these managers have a bias to being short of volatility,” explains Merle. 

Discretionary inputs can include politics; regulation; weather, other fundamentals, and almost anything heard “through the grapevine” from cash traders.

Though some systematic managers have codified weather signals, for Merle weather patterns require discretionary analysis. He explains: “For natural gas there might be a fairly simple linear relationship between temperature and demand, but for other commodities the relationship is much more nuanced. For grains, either not enough rainfall or too much rainfall can be bad, and it depends on how close the timing is to the harvest. Extreme weather events, defined as three standard deviation events, are now 400 times more common than between 1951 and 1961, according to research by the Goddard Institute of Space Studies. Climate change is not only increasing temperature but also the standard deviation of temperature, which produces much more extreme events”.

Judgment has led Merle to scale back some soybean trades due to uncertainties around the US-China trade war and associated tariffs. “Tariffs are a structural change. Therefore, we have reduced risk and volatility budgets for soybeans because the models cannot see this kind of structural change,” says Merle. He also keeps a close eye on soybean markets and currencies in Brazil and Argentina, which together produce more soy than the US (although he does not trade any South American soybeans). 

Discretion holds more sway over initiating trades than closing them. The quant model signals may only account for 20-25% of trade entries, but make up more than 80% of trade exits. Trades can be closed if they are approaching the pre-set stop loss; if the portfolio volatility target is threatened; if correlation between trades has breached 30%, or simply if the models, which update in real time, have started forecasting losses from the trades.

“I need to understand how and where corn grows, where the flows are, how grains are traded around the Black Sea, which is important for wheat and a strong regional player for corn, and the issues cash traders are considering,” says Merle.

Repertoire of trade types

CTAs trade many agricultural commodities but generally on a directional basis, whereas Merle is a relative value trader. His largest trade type, and source of profits so far, has been term structure or calendar spread trades, making up about 60% of exposure. These will typically wager on the steepening or flattening of a curve. Merle also reckons his term structure models are a bit better than his other types. The tenor of these trades varies according to market liquidity: he might go out as far as 18 months for grains, but could trade a contract 3 or 4 years ahead for energy, in order to avoid becoming too large a share of open interest.

In March 2020, trades anticipating a steepening of the term structure for corn proved to be profitable. In April 2020, a curve trade structured to profit from a steepening curve in WTI Oil, actually profited from negative prices, but has anyway been switched into an option structure at CME’s Nymex Clearport.

Cross market trades – half of which are also location spreads – have been the second largest profit driver. They are a smaller bucket partly because they generally do not take place on the same exchange and are therefore more margin intensive absent netting between exchanges. (Incidentally, Merle could be interested in brokers who can offer such netting). Such trades can also be logistically more complicated to execute: while some exchanges offer ready-made spread trades, where legs of a trade have to be executed separately there is more risk of prices moving adversely before the full structure is put on. To address this risk, Merle is developing algorithmic execution capabilities to allow for simultaneous execution, and this may increase the potential allocation to cross market trades. Similarly, butterfly spreads can come prepackaged on CME/CBOT/NYMEX but would need to be constructed from their constituents on Euronext and ICE, which reduces the current budget for them.

The third largest category – product trades – are based on relative pricing within the production chain between raw material commodities and refined products, such as soybeans relative to soybean meal and soybean oil.

Merle does not trade option spreads but uses options for tail risk management.

Climate change is not only increasing temperature but also the standard deviation of temperature, which produces much more extreme events.

Olivier Merle, Founder, Blackbird Asset Management

Risk management

The strategy is long of out of the money call options as a tail risk protection overlay. Notwithstanding the plunge of crude oil into negative territory in 2020, commodity prices have historically tended to have their sharpest spikes upwards, on supply panics. These options did not produce profits in March 2020, because prices were trending down.

Risk management has several other layers. The 1x leveraged strategy targets 7.5% volatility and the 2.5x leveraged version (where most assets reside) targets 18.75% volatility. Realised volatility has been in line with the target about 90% of the time, though Merle left some headroom for an uptick in volatility to around 10% in order to take advantage of opportunities in March 2020. He only feels comfortable increasing the risk budget in this way after a winning streak; the strategy volatility also perked up in May 2019 when the call options were moving closer to their strike prices.

There are also trade stop losses. “The target is to avoid losing more than 1.5% per trade, though occasionally losses have been larger, around 1.9%, due to gap risk rather than slippage. Liquidity related transaction costs related to bid/offer spreads did widen in March 2020, and currently form part of the discretionary risk overlay (but should be codified as part of the algorithmic execution project),” explains Merle. 

Margin to equity for the 1x leveraged strategy is capped at around 35%, but has averaged around 20%. Margin rates can sometimes be increased by exchanges and/or brokers without prior notice. They did increase in March, but by less than the increase in volatility. Currency risk is hedged. 

Service providers

“INTL is one of the biggest non-bank FCMs (Futures Commission Merchants), specialising in clients who are cash traders,” Merle says. Merle is happy with INTL FC Stones’ competitive pricing on execution, though he is looking to add another FCM. 

Exchange fees around 2% per year are a much larger cost than brokerage commissions, so going forward any way to access exchange memberships could reduce costs. Exchange fees are relatively high because trading 9,000 round turns per US$ million per year means the strategy has nearly 20 times more trading than a typical long-term trend follower, which might trade just 500 round turns per US$ million. 

“Administrator Sudrania provides excellent fund accounting, fee calculation, administration and investor reporting at very honest prices, with double validation. They also assist with some of the NFA (National Futures Association) reporting,” says Merle. BMO Harris is custodian.

Why raise assets from the US?

Raising tens of millions in two years is no mean feat for a startup, and innovative European managers may need to search outside Europe for seed funding. Although Merle sits in Paris, the investors mostly come from the US. This is not entirely surprising, since roughly two thirds of hedge fund industry assets do come from the US. There are also special reasons why his strategy has transatlantic appeal. “Some European investors cannot invest in commodities. US investors understand the asset class better, and they are also more comfortable with managed accounts,” he explains. Additionally, UCITS is more popular in Europe, but Merle has no plans to introduce a UCITS, given the complications entailed in obtaining direct commodity exposure.

Merle’s first third party allocation came in March 2017 via a managed account hosted by RCube Asset Management, which is an emerging manager incubator platform taking care of the risk management, back and middle office, regulation, compliance, marketing and eventually asset raising, so that Merle can focus on research and trading. Assets of US$68 million are roughly split between the managed account (the RCube Commodity Spread Program), which contains a US pension fund, and the US commodity pool structured as a Delaware LLC, BlackBird Alpha Spread (Legal name: BlackBird Alpha Series 1: BlackBird Alpha Spread 1x & 2.5x), which runs pari passu (though the managed account uses the same FCM, it gives up trades to another FCM). Merle remains the second largest investor in the strategy and pays the same costs as any other investors; he shares the legal, set up, bank fees with other investors but does not pay management or performance fees. Setting up a fund structure has reduced exchange fees by about 50%. 

As of May 2020, the strategy is closed. If and when it reopens, the current regulatory structure means that it could only raise money from European investors via reverse solicitation. As Merle has most of his net worth in the fund, he is prioritizing returns over asset gathering. Reopening partly depends on how new strategies and execution methods pan out.

Future research 

Merle is steadily building out energy and metals, following the tested pattern of piloting trading in his personal account before adding new markets to the fund. He traded energy for a year or so in his personal account, before allocating 10-20% of the fund’s risk to it, mainly in term structure trades. Merle is looking to hire an energy analyst to further grow this area. He will also soon begin trading precious metal term structure, and is contemplating adding iron ore.

The investment universe could also be expanded by trading on certain exchanges in China and India that require special permissions.

Though there is no hard target for capacity now, it could increase over time as the investment universe is expanded and as algorithmic execution, which may minimize market impact, is introduced. Merle is cautious on capacity, because his years of trading have taught him that it can only be discovered through experience.

The execution project entails one full-time employee and one consultant combing through the huge and expensive tick datasets. This research is not purely for trade execution, it could also lead to a short term purely quantitative strategy being developed within the fund. Merle’s appetite for innovation is unabated.