On June 14,TellusLabs will be attending the BattleFin Alternative Data Discovery Day on the USS Intrepid. The event is the perfect place to continue our dialog with the investing community about how a new generation of satellite imagery-based insights is poised to have profound impact across a surprisingly wide swath of asset classes. Our CEO, David will be speaking on a panel there and as a former Navy officer himself he’s particularly excited about the setting.
BattleFin is the latest in a set of “FinTech” events that TellusLabs has participated in over the last few months, including StockTwits’ Stocktoberfest East (where our Head of Product, Greg Allen, was converted into a RobinHood addict), and FinTech Sandbox’ Demo Day where we had the opportunity to present TellusLabs to Boston’s financial community.
At these meetings, and in our customer discussions, we’ve picked up on three themes in FinTech today that get us excited:
The financial community is embracing the power of novel data in a big way, so it’s natural that today’s ‘exotic’ data providers are tracking a rapidly expanding universe of ‘data exhaust’, including: footfall in retail stores, AIS from freighters, mentions on social media outlets, cell phone signals, credit card transactions, and micro-scale weather.
Given how much of the economy plays out in plain sight around the globe, could there be a more compelling class of data than satellite imagery? NASA's MODIS satellites have been orbiting the planet for over 15 years, taking 4 pictures a day of every square mile on Earth. There are some pretty powerful stories to be told and some serious alpha to be unleashed, but delivering on the potential is not easy... which brings us to the second theme we’re excited about.
New and differentiated datasets wield enormous potential, but that doesn’t make them valuable. Big Data only becomes useful once it’s been converted into insights that can fuel better decisions. The game-changer for the finance sector isn’t necessarily the ubiquity and diversity of the data available - though that’s certainly a prerequisite - it’s the technology and expertise to harness the data and answer the real underlying questions. Doing that involves superior compute and mature machine learning methods but also domain expertise and savvy, experience-based feature engineering.
One of the things we love about the finance sector (and the Ag sector, for that matter) is that the market participants are remarkably results-driven. We happen to think satellites and Earth observation data are really cool... but we understand our customers don’t necessarily share that passion. We earn their respect by delivering results. Last year, our agricultural intelligence product, Kernel, predicted end-of-season US Soy yield with 0% error (and Corn within 1%). For Soy, we were within 1% of the right answer two months before the USDA’s own forecasts. But being right is not always enough, and that’s our third and final observation…
We think it’s fair to say tech products for the finance industry haven’t always had a lot of empathy for the user. On the retail side, we’re all acquainted with clunky interfaces and convoluted fine print that make it harder, not easier, to participate in the financial markets. Even on the enterprise side, the world’s elite institutions often accept market standard technology products that are aesthetically unappealing as well as difficult to navigate.
That approach is changing. Consumer apps like RobinHood, Coinbase and Mint are beautiful and intuitive. Enterprise/SMB solutions like Quovo or Ripple also deliver complex products to important institutions in a clear and elegant way.
Kernel is informed by the same philosophy. Through Lykaio and Greg’s immersive web portal, users can see confidence intervals and historical comparisons for our forecasts, drill down to individual states and even counties, and they can see how much each forecast is influenced by satellite data, weather data or ground observations. Our design aims to be simple and intuitive, but to also incorporate the beautiful imagery that serves as the foundation for our modelling work.
We also know that ‘last mile’ means more than an application. Research functions within financial institutions are well staffed with in-house data scientists and engineers whose task is to use disparate and novel data to give their firms an edge. For us to help them get the most from Kernel, it means getting our insights (and much of our data) to them on their own terms, via a modern, performant API.
These three themes will take time to play out, as the alternative/‘exotic’ data landscape in FinTech takes on a more mature role in the ecosystem. At TellusLabs, we’re looking forward to continuing this dialogue at the BattleFin Data Day on the Intrepid (and beyond)!