Chris Kantos: How Natural Language Processing Can Generate Alpha

Can you analyze social media for investment decisions? How do hedge funds use Reddit posts, earnings calls, and SEC filings to find alpha? What’s the role of LLMs for financial analysis in 2025? Chris Kantos, Head of Quantitative Research at Alexandria Technology, joins us to explain how the buy side uses natural language processing (NLP), AI for investing, and text-based sentiment data to generate AI alpha signals across all asset classes—from equities to commodities.We dive deep into how Alexandria builds quantitative trading strategies from unstructured data like Reddit posts, news articles, earnings calls, and 10-Ks. Chris explains how most hedge funds get NLP wrong, why Alexandria’s document-specific classifiers give them an edge, and what makes a good social media for stock analysis dataset in a crowded and noisy world. He also tackles the myth of data commoditization and explains why alpha decay isn’t always inevitable.We answer questions like:– How do hedge funds use Reddit and social sentiment in trading?– What makes a good NLP model for financial data?– Has alternative data become commoditized?– What separates FinBERT from other finance-specific LLMs?– What’s the best way to train a sentiment model for earnings calls?– How is ChatGPT used in finance and investing workflows today?– How do professional quants cut through the noise on Twitter, Reddit, and X?– What’s the future of AI in systematic investing?– How does news sentiment impact trading strategies?– How do hedge funds use alternative data beyond equity alpha?– How do professionals use Reddit data for stock analysis without getting fooled by noise?– What’s the best path to become a quant researcher today?– What skills and experience matter in quant finance careers?Chris also shares quant finance career advice, why he left risk modeling for alt data, and what really happened in the office when Bernie Madoff got caught.