According to a recent study by McKinsey & Company, commodity trading cash flow almost doubled between 2018 and 2021. The study also found that while the commodities market has continued to grow over the last two years, it has also been subject to fluctuations caused by four major factors — including extreme weather.

Moreover, the study concluded that the majority of players in the market have been hampered in their ability to quickly respond to these growing challenges. The reasons for this include outdated technology infrastructure and a shortage of “commercial talent with experience in data-driven methodologies.”

Reading between the lines

Severe weather often plays a key role in disrupting the commodities market. Heat waves, droughts, heavy precipitation, and other high-intensity weather events can abruptly alter supply and demand, which can have an immediate impact on availability and pricing of soft commodities (e.g., agricultural products, livestock) and hard commodities (e.g., oil, natural gas, rubber). A commodities trader who could improve her “batting average” by using more accurate and reliable subseasonal-to-seasonal (S2S) forecasts 2-weeks to 1 year ahead would enjoy a clear competitive advantage.

Many commodity traders have an illusion of competitive advantage regarding S2S forecasts. Some firms dedicate impressive resources to the problem, with teams of 20 or more full-time meteorologists. Other firms outsource their meteorology capability and-or subscribe to commercial weather vendors, who specialize in short-term forecasting, and include S2S data as an add-on.

On both ends of the spectrum, what distinguishes the resulting S2S data feed is not the underlying model itself, but largely how the information is packaged. The foundational data originates primarily from two sources: National Oceanic and Atmospheric Administration (NOAA) in the United States and the European Centre for Medium-Range Weather Forecasts (ECMWF). 

Most commodities traders live with this status quo. As one trader told us, legacy S2S forecasters “all say the same thing, and no one goes out on a limb.”

Finding a difference-maker

Other than some simple bias-correction, there’s not much S2S data providers can do to differentiate a weather forecast based on data furnished by government agencies whose models use known scientific and statistical connections based on historical patterns.

At Salient, our machine-learning experts apply artificial intelligence to find unknown, lesser explored, often non-linear scientific connections (involving such variables as sea surface salinity) along with the more established ones. The result is an end-to-end validated, calibrated, probabilistic AI model with non-consensus views.

We also make adjustments that account for uncalibrated data in the government forecasts. Say, for example, a government forecast showed a 60% likelihood of an extreme drought. If reviewing the historical record reveals that 60% likely forecasts for extreme droughts only occurred 40% of the time, that government model is statistically overconfident. At Salient, we adjust those probabilities based on actual results to make our S2S forecast system more reliable, and that means more confident trades.  

Playing the percentages

The implications for commodities traders are clear. If you trade in oats or barley, the difference between a 60% and 40% forecast of drought in a region that produces a significant amount of oats could have a huge impact on your decision making. Given the timing and severity of a drought, it may stress the crop resulting in a lower yield, depending on additional variables such as the seed type and the crop growth stage. Lower yield means lower supply, which means higher prices.

Salient’s system allows traders to monitor any number of regions as our model updates, indicating shifting weather probabilities. Our model also shows the likely impact on different commodities (e.g., corn, soy, natural gas) because each has different thresholds (e.g., precipitation, temperature, growing degree days) that can be defined in the system.

Find a winning strategy

As the McKinsey study indicates, commodities traders who want an advantage must upgrade the technology they use and adopt data-driven methodologies and research. Improved S2S forecasting technology that leverages machine learning is an obvious place to start. Salient’s goal isn’t just to be different — it’s to be better.

In such a highly competitive field, any edge can make a difference. If a commodities trading firm makes 100 weather-based decisions a day, an accuracy and reliability gain of 5% using our forecasts over other baseline models compounds over time. Not every commodities deal has to be a home run. Success is more about improving your batting average.

In baseball, what’s the difference between an elite hitter’s .300 batting average and a mediocre hitter’s .250 batting average?

Five extra hits for every 100 at-bats.