As the world of artificial intelligence expands, so, too, do the opportunities to understand digital data and to gain market advantage. A new TABB Group research piece breaks down market information collection methods such as satellite intelligence, drone piloting, Internet data harvesting and market transactional data collection as it points professional investors to a new level of informational access in the digital age.
Data Collection, Photo by olafpictures, Pixabay
Mining the Digital Age for Wall Street Intelligence
In an age of data tracking people’s every movement, global intelligence agencies collecting and storing phone calls, text messages and emails, companies “are increasingly able to produce cheaper satellites, sensors, and drones, which capture different types of data from high in the atmosphere,” TABB Group’s Valerie Bogard noted in the November report titled “Capturing Alternative Alpha: The Rise of Alternative Data Sets.”
But it is not just in the air that information is increasingly being procured. On the ground, information is being mined from smartphones and mobile devices and downloading an ever-expanding list of applications.
Related article: The Crowd Mine
At first, these technologies were used primarily as a tool to deliver a particular function, such as make phone calls or connect to the Internet. But now there is a growing data analysis opportunity that is being spun off from technology’s core functionality.
“While active managers are facing unprecedented competition from exchange-traded funds, fund managers are also struggling to find alpha from their traditional data sources,” Bogard said in a statement. “Firms are gradually realizing that discovering new trading strategies also requires the discovery of new data and differentiated insights, which is driving a deeper hunger for alternative data as the answer.”
Four Categories of Data Collection
TABB’s research divides the market data opportunities into four categories: Public Website Data, Individual Location Tracking, Satellites or Overhead Surveillance and Consumer Transactional Data.
The application that has received the most media attention is satellite tracking, which includes drone data collection, is most commonly used for car counting exercises to determine economic activity. But these uses are beginning to expand. Satellite images can estimate factory employment, track oil tankers to determine supplies, the height of buildings in a growing Chinese city can provide clues to economic activity, while stockpiles of metal and materials can help a market analyst understand commodity supply and demand.
“Making sense of geospatial data may seem easy to a human eye, but it takes a combination of complex artificial intelligence technology and processing power to recreate this ability in machines, especially when it must be done quickly and across millions of images,” Bogard wrote.
You may also like AI Advances Hint at Neural Networking Future
Numerous data providers provide the analytics and they team with satellite providers, many of whom are launching smaller “nanosatellites” that are increasingly focused on a narrow list of tasks. The industry is currently working on developing stronger correlations between the imagery it captures and economic meaning.
Consumer transactions are being analyzed at an increasingly granular level by various companies that provide clues to economic trends. These efforts are being met with increasing data structure challenges and varying privacy rules.
Location data captured by social media properties such as Foursquare are combined with cell phone location tracking. The user’s demographics can be matched with their traffic patterns to determine supply and demand trends. “This data can then be used by traders to gain insight about power consumption and how it relates to commodity and energy supply chains,” for example. “Whether these location tracking firms are generating their own proprietary sensor technology or inputting data from apps and other sources, making sense of spatial data requires a high level of unique analysis and validation techniques to make it useable.”
Check out One Algorithm to Rule Them All
Public website data is used in a similar vein to location tracking, but the signal to noise ratio is typically problematic. “The majority of data extracted from the Internet is not useful for investment decision making so the first step for any alternative data provider is to try to filter out some of this excess data,” the report noted.
The market for these services is large and growing. TABB Group estimates the US alternative data market to be at 0 million, which could grow to 0 million in 5 years. “As companies increasingly become data savvy and seek to monetize their exhaust data, this base is also growing immensely.”
By Mark Melin
Read next: Quantum Computing Takes Flight
