New FinTech Course Explores Language-Based AI
Finance may be an industry dominated by numbers but David Pope ’90 believes that words hold the key to creating high-performing portfolios.
Investors are increasingly using AI tools to inform their financial decisions, the lecturer in Finance (and Bentley graduate) explains. He notes that natural language processing (NLP) programs, which recognize and interpret patterns in human speech and text, are proving particularly adept at transforming massive sets of data — gleaned from earnings call transcripts, annual reports, acquisition announcements and other finance-related corporate communications — into actionable insights.
This fall, Pope teamed up with Bentley colleague Tamara Babaian, a professor of Computer Information Systems (CIS), to introduce “Investment Applications of Natural Language Processing” (FT 370), an interdisciplinary course offering students a comprehensive look at the theoretical foundations and practical applications of NLP programs in the finance industry. A required course for students majoring in Financial Technologies (FinTech), it features weekly lab-based sessions that teach students how to create their own machine-learning models using Python, a popular programming language.
According to Pope, these hands-on coding sessions distinguish Bentley from other universities offering FinTech courses. “Having firsthand experience coding NLP programs gives our graduates a deeper understanding of the technology and its applications, and thus a competitive edge,” he explains, noting that the cutting-edge information covered in class is “so new it’s not even in textbooks yet.” Instead, he offers students access to newly released research papers as well as personal and professional insights: Founder and CEO of Speech Craft Analytics — a company using AI to analyze “vocal features that are imperceptible to the human ear” that reveal a speaker’s “true emotional state and intentions hidden within their words” — Pope has more than 30 years’ experience in financial and data analysis.
For the course’s culminating project, he and Babaian tasked students with building their own NLP programs utilizing the Gunning Fog Index (GF), a measure of language complexity based on sentence length and the number of syllables per word. “Executives who use complex language on earnings calls typically do so to obfuscate or delay the release of bad news,” Pope explains, “and as a result, their companies’ stocks underperform.” Based on their programs’ findings, students created portfolios with stocks purchased from companies with lower GF ratings (e.g., simpler language on their earnings calls) — portfolios that “would have outperformed the S&P 1500 by 3.8% each year from January of 2011 to June of 2023,” he observes.
