Using Satellites to Forecast Metals and Commodity Prices
You can be a tech firm with the most innovative product on the planet, but the only way you’re going to sell that product is if people know about it. You can spend loads of money paying some mediocre PR firm to cold-call 1,000 different media outlets with a “pitch” – essentially a request for journalists to do free work – or you can try to spread the message yourself on platforms like Medium.
It’s a shame that most the content on Medium is inane drivel, because there are some real valuable articles that get passed by for promotion in favor of hard-hitting thought leadership pieces like “I Met a Nazi on My Vacation.” We occasionally come across good Medium articles while doing research, but this time, one was sent to us by the Chairman and CEO of RS Metrics, Maneesh Sagar, who thought we ought to take a look at the work his company is doing with using satellite imagery to forecast the supply of metals. It’s nothing short of fascinating, so let’s dig in.
We first came across RS Metrics in our article on 8 Satellite Data Startups Doing Geospatial Analysis where we briefly talked about how they use satellite data for business intelligence. They’ve since started to productize their offering into four main areas:
We’ve talked before about how firms are using satellites to count cars in store parking lots, so nothing really new to talk about there. We’ve also looked recently at how satellite imagery is being used for commercial property valuations. As for Tesla, monitoring the number of Tesla vehicles being produced or keeping an eye on Elon Musk’s receding hairline both sound cool, they’re pretty limited use cases. The product we want to talk about is Metal Signals.
Before we dig into what Metal Signals does, we first need to understand the basic use case which involves going back to Economics 101 and thinking about supply and demand in the context of commodities. Most retail investors are completely oblivious to the world of commodities, where traders use real-time newsfeeds to gauge the impact of real-world events on the supply and demand for commodities. Every bit of information can be used as an advantage when it comes to trading, but only for a very short period of time. There’s actually an entire theory around this idea – something you MBAs should be quite familiar with – called the efficient market hypothesis. We can see your eyes glazing over, so let’s get back to talking about how we can make money with commodities.
Supply and Demand
In much the same way that you can trade stocks, you can also trade commodities. Consequently, any additional information you can use for trading, that may not be so widely available, gives you an information advantage. Since commodities move in response to fluctuations in supply and demand, being able to predict supply or demand gives the trader an advantage. Predicting supply seems a whole lot easier because that means looking at all the metal laying around waiting to be sold. RS Metrics describes this as follows:
Historically, base metal prices maintained a strong relationship to the change in inventory of metal stocks stored at exchange-approved warehouses used by the London Metals Exchange (LME) or Chicago Mercantile Exchange (CME). These are commonly known as “on-warrant” storage facilities.
In other words, just measure the inventory of all metal stored in the warehouses that store metal and there you have it. The problem is, not all metal stockpiles are being stored in these approved warehouses.
This is where things start to get really interesting. RS Metrics has developed a solution that actively monitors more than 400 global smelters (places where ore is turned into metal), terminals, and storage facilities, to see just how much metal is actually being produced and stored. Here’s an example of nearly 850,000 tons of aluminum stored “off-warrant” – in other words, under the radar – in an Aluminicaste facility in San José Iturbide, Mexico.
RS Metrics uses satellite imagery to capture changes in metal inventories at “off-warrant” storage facilities like the one seen above, transfer points where metal is stored between destinations, and the actual production of aluminum, copper, and zinc at the smelters themselves in near real time. Below you can see an example of how aluminum ingots and alloys can be captured from a satellite image.
They don’t just stop at counting the metal ingots, they go much further than that. Let’s take a look at copper in our next example.
Monitoring Copper Production
When it comes to monitoring production, RS Metrics does some pretty cool stuff like tracking the movements of vehicles and dump trucks which can help gauge overall production activity. They can even count the number of people that showed up for work based on the cars in the employee parking lot. It’s remarkably responsive, since more than 90% of their imagery is 48 hours old, and they can actually sample areas of interest they identify at a much greater frequency than that. (They even have historical data going back to 2013 – which is super important for the Quants – who have to back-test everything. How cool is that?) When it comes to copper, understanding the production process a bit helps you understand how they’re able to predict production at a smelter. Check out the below diagram:
As you can see, various steps involve concentrates, anodes, and cathodes. While measuring concentrates from the air can be challenging, RS Metrics can measure stockpiles of anodes and cathodes using a patented process. These are the sorts of insights that help make their predictions far more accurate. All this data can be queried through an application that lets you drill down to specific outputs by specific mines. Here’s an example of monthly cathode measurements from the Escondida copper mine in the Atacama Desert in Northern Chile, currently the world’s largest copper mine by reserves.
Aggregated weekly or monthly data is available for out-of-the-box analysis, and they use Kalman Smoothing – some cool-sounding mathematical technique we know absolutely nothing about – to adjust for missing values. All of this data can then be purchased via subscription for around $1,200 a month. In the world of financial data, that’s actually a pretty good price, and once you’re a subscriber you can then start using that data to generate alpha.
The term “generating alpha” is mainly used by pretentious pricks in the finance world who want to sound self-important, and consequently, we love talking about “generating alpha” every chance we get. In layman’s terms, “generating alpha” means capturing profits off trading on information that other people might not have, just as we were discussing earlier. The RS Metrics research team has found that their Metal Signals data has a statistically significant correlation with directional changes in commodity futures, metals-related equities, indices, currencies/FX, and interest rates. For example, just a few days ago Bloomberg was talking about how a quarter of South Africa’s currency moves can be ascribed to iron ore prices. It’s incredible to think just how closely connected our small world is.
So how well does this stuff work? Pretty darn well. In a sample report RS Metrics sent us, they demonstrated how their signals could be used to generate superior performance in trading currencies, commodities, and mining stocks over a four-year time frame. Using aluminium as an example, Metal Signals aluminum signals have been 70%-80% predictive of LME futures price direction 1, 2, and 3 months out, depending on which forecast model is used. In the below example, we can see how President Trump’s tariffs against China impacted aluminium production in China based on some spies in the sky.
We’re always skeptical about applying these sorts of signals to ETFs, but the management team at RS Metrics showed us come compelling collateral which demonstrates how they might be able to create some viable financial products around this interesting technology. They’re in the process of putting together an ETF strategy around Metal Signals so we’ll look to cover that in a future article. Stay tuned.
While some Medium users might prefer to read riveting real-world accounts such as What It’s Like to Be A Woman On the Internet, we think they ought to give the RS Metrics article a read instead. It’s a fascinating look into how satellite data can be used to glean insights that help us predict supply and demand for some of the world’s biggest commodities. If you’re a company like Glencore, you pretty much have to subscribe to this data product. Heck, if you’re a company like Glencore, you might forgo subscribing and just try to buy the entire firm. It’s pretty easy to see that RS Metrics will have no problem finding suitors when it comes time for a liquidity event. If you’re interested in purchasing a subscription while you still can, just drop them a line.
Pure-play disruptive tech stocks are not only hard to find, but investing in them is risky business. That's why we created “The Nanalyze Disruptive Tech Portfolio Report,” which lists 20 disruptive tech stocks we love so much we’ve invested in them ourselves. Find out which tech stocks we love, like, and avoid in this special report, now available for all Nanalyze Premium annual subscribers.