Ben Wellington – Complex Feature Engineering at Two Sigma (S7E32)

Feature engineering serves as the critical middle layer in modern quantitative finance, transforming raw data into economically meaningful signals that drive alpha. As raw data becomes increasingly commoditized, the primary edge shifts toward creative, human-led feature generation that identifies unique, orthogonal patterns. The emergence of Large Language Models (LLMs) has fundamentally altered this landscape, enabling researchers to rapidly prototype complex ideas—such as analyzing CEO micro-expressions—by treating diverse data types as language. While these tools lower the barrier to entry and accelerate research, they also risk reducing signal entropy and increasing model correlation. Consequently, successful quantitative firms must leverage AI as an amplification tool for human intuition, prioritizing diverse, idiosyncratic insights over simple automation to maintain a competitive advantage in an increasingly efficient market.