LLMs for Equities Feature Forecasting at Two Sigma with Ben Wellington - #736

In this episode of The TWIML AI Podcast, host Sam Charrington interviews Ben Wellington, Deputy Head of Feature Forecasting at Two Sigma, about how LLMs and GenAI are changing problem-solving approaches in the financial sector. Wellington discusses his background in NLP and how Two Sigma uses features to predict the future prices of tradeable instruments. He explains the process of identifying and collecting data for features, the importance of historical data, and how LLMs have significantly reduced the time and resources required to create and analyze features. The conversation covers the evolution from one-hot encoding to embeddings, the challenges of noisy signals in financial modeling, and the potential of agentic models. Wellington also touches on the build-versus-buy decision-making process, the importance of open-source models, and strategies for preventing data leakage.