To stay up-to-date with the ever-changing technology industry, CME Group hosted Tech Talk, an event to bring together industry leaders to highlight disruptive financial technology and how it will impact the financial markets.

David Gresky, head of market technology sales, Americas, at CME Group, moderated a panel with David Potere, CEO and co-founder of TellusLabs, and John Han, head of research and analysis at Orbital Insights. They are two leading alternative data providers specializing in satellite imagery. 

The panel discussed alternative data and its effect on capital markets and how the data ties into the agriculture and energy markets.

Here, we share the highlights and excerpts edited for MODERN TRADER from the Tech Talk 6.0 event, which was held in Chicago this past summer. 

David Gresky: How did you get involved in satellite Data?

David Potere: I met my co-founder in 2003, Professor Mark Friedl, at the beginning of my Ph.D. program in geo-demography at Princeton. I fell in love with satellite imagery as a kid using NASA and Navy imagery for mine hunting. Mark and I were convinced from the beginning that there are practical non-science applications for this data from all the satellite constellations.

John Han: Orbital Insights is a relatively new company, founded in 2013, but we found that there was huge demand for satellite data because it really does not exist at scale today. Counting cars and measuring agricultural yields and looking at ships being built in ports is hard data to find. There is a lot of demand from the financial markets and the corporate markets. Everyone cares about two important things — increasing revenues and decreasing costs — and this data helps do that from a corporate and investment side. 

Gresky: What are you seeing with your customer base and how are they using this technology today? What are the most interesting applications you are witnessing?

Potere: Our flagship product is called Kernel, which is a software application that’s 100% focused on agriculture. We monitor the 180 million acres of corn and soybean yields in the United States. The folks we work with tend to value discretion, but I can share that it’s been a very collaborative process. We are constantly learning. We are an [application programming interface]-first company, because we learned early on how sophisticated the data science and analytics teams are at our client’s institutions. Clients like that we have a web app and front end that helps them explore and understand the data, but most data science teams just say, “give me the API key.”

Clients typically pull our data into internal systems for support and validation for in-house models. So, we need to become educators and evangelists around the data.

While the prediction or signal is the most requested part of the data, often data science teams want to analyze other raw data and take steps to validate their view.

Han: On a similar level, we think about how clients want to use our data; some want the raw data, other clients want normalized data to compare on a relative basis, and some just want the conclusion or signal to buy or sell. We can tailor the data to a client’s needs.

For example, some clients want global data on crude oil tankers around the world, or we can get more granular on how much inventory is in the various OPEC countries or regions.

Gresky: Satellite technology has been around for a long time. Why now? What are the main drivers making satellite data so popular? 

Potere: One driver is that satellites have been around long enough for the data to have a good record to train against (every season they see improves their understanding). We now have enough meaningful historical data points to take a stab at researching this in a meaningful way. The U.S. government has been archiving images through Aqua & Terra satellite projects four times a day for the last 16 years. That’s a lot of variation of data that provides adequate data to validate research ideas on El Niño, La Niña and many other events. And the computer technology has improved dramatically, so we can actually run modern machine learning on data at this scale. If it wasn’t for AWS Cloud Services, we would have had to partner with an institution that could provide that resource.

Han: It’s a combination of three major drivers.  First, the cost of building and launching satellites is coming down, and this increases the number of satellites and different types of data collected. Second, there now is enough data to efficiently train neural networks. You can’t train a neural network with five photos, but when you have thousands and thousands of photos training the data, you get insights. Neural nets can teach the data that this is an image of a mill in China or this is a ship; it can understand what it looks like, then your [artificial intelligence] works. Third, new cloud storage services. Previously you would have spent millions of dollars to store data, but now data providers can store and process this data daily and with a small budget. 

Right now, for example, we are monitoring 250,000 parking lots in the United States, and in the whole world we monitor every single oil tank that’s floating every single day. So, daily we can tell you how many cars are in a parking lot and how much oil is in a country. That technology did not exist 10 years ago.

Gresky: What do you see in the next five years? Where do you see technology progressing?

Potere: Five to 10 years in the future, you’ll see data analysis become a crucial skill throughout the economy, and we’ll have more analysts than ever in the workforce. These 21st-century professions will be solving more problems than ever with geospatial data. They’ll need tools to help them solve real problems.

Han: When we think about what happens in the next 10 years, the cost of satellites and hardware will be cheaper. There will be more data and hot air balloons and drones flying around collecting data 24 hours a day. The bottom line is there will be a huge increase in data and that data will need to be analyzed by somebody.

In terms of the applications, we are finding it’s still an education process. We are exploring new questions every day. For example, how does Amazon Fresh affect other grocery chains? Are Chinese steel mills operational? And how does it affect iron ore miners using infrared satellite data? You can ask great questions that previously you would not dare to ask. 

Gresky: How can exchanges like CME Group help in awareness and application of such data?

Potere: The market has gotten used to branded indexes. Great monitoring systems with value-add analytical layers can form the basis of new indexes. These indexes can provide transparency, and that’s a role exchanges can support. 

Han: I agree that indexes will be a huge part of the role of exchanges. The data is so objective and can be provided daily. Because of the objectiveness, cadence, and granularity it’s a superior index product. I’m very confident that there’s a lot of indexes and new products that could be created by exchanges.

Gresky: What’s one parting note that you want people to know about satellite imagery?

Potere: That’s a challenging question. To put it in perspective, we have only touched on [a tiny fraction] of 1% of the data collected so far, and data is increasing on an exponential curve. What happens when someone touches the first page of the 500-page photo album of the earth, and then what happens when they can slice through the entire photo album. We are right on the cusp of putting the whole photo album in flight. 

Han: The satellite imagery application is so big; we still have not nearly reached near the potential of artificial intelligence technology. If you are looking at Google Earth or Google Map and you see use cases that are interesting, then we could probably provide insight.