Alternative Data for FinTech and Business Intelligence
Lin William Cong - Cornell University;
Beibei Li- Carnegie Mellon University;
Qingquan Tony Zhang - University of Illinois, Urbana Champaign;
-- The value of alternative data is gaining much attention in recent years; asset managers from all walks of the field start to realize the importance of such information. Our recent article, "Alternative Data in FinTech and Business Intelligence," introduces some of the latest research in economics and business-related fields utilizing data from unconventional sources or of unstructured nature. The paper highlights unifying themes of such big data and the methodologies for analyzing them at scale.
Our article first elaborates the applications of textual analysis along with images, voice and video in corporate finance, investment, and macroeconomic forecasts. As the most salient alternative data used, earlier studies using textual data are typically count-based and rely heavily on researchers to pre-define a relevant dictionary or word list. As machine learning becomes more frequently used to process textual analysis, more unsupervised learning tools have appeared. In the section on image analysis, the article discusses the application of satellite images and the use of profile photos in financial industry; interestingly, the appearance of fund managers tends to have a relationship with their performance. The research of voice and video is still at an early stage compared to the research of text and images, but valid outputs have been revealed as well. Some find that the emotion and tone in customers’ voice contributes to identifying their traits and personalities.
Another amusing topic that we discuss in the article is the digital footprints from social media and mobile devices. The rise of social media and the common use of mobile devices have reshaped the structure of information collected from these platforms. Digital footprints can be generally defined as the new and unparalleled sources of fine-grained user-behavior data generated from social media and mobile devices such as individuals’ cellphone usage, online and mobile activities (e.g., web browsing, click-stream and tap-stream, shopping and payment), social media and social network activities, GPS locations and movement trajectories. The article takes financial credit risk assessment as an example and discusses how such new sources of alternative data can be leveraged to improve financial predictions, profitability, and social welfare; as well as limitations and recent progress.
Another major source of alternative data is from Internet of Things, which has empowered almost every industry to become much more efficient. IoT has improved customer experience and optimized supply chain operations in many ways, which had a huge impact on retail industry. Data from IoT can be categorized based on the properties of the sensor, such as geolocation data, which our article examines in a few cases in detail to illustrate such topic, as well as image-based IoT applications. The geolocation, image and transaction data streams from over 400 retailers and distributors have only been part of the alternative data that have been utilized. Within five years, the consensus view is that IoT data will become the largest volume of alternative data for finance analytics. Both new challenges and opportunities will emerge as more dynamic and advanced IoT devices are developed. Lastly, the article expressed the need for future research on alternative data and a positive attitude on the rise of such information to be used in the world of FinTech and business intelligence.
Overall the article offers new perspectives on the novelty of the use of alternative data in the financial industry, which are worth studying and conducting further research on.