Macro Hive Conversations With Bilal Hafeez Ep. 312: Deepak Gurnani on Five Steps for Using AI in Investing
55 snips
Jun 12, 2025 Deepak Gurnani, Founder and Managing Partner of Versor Investments and former CIO of Investcorp, shares insights on integrating AI into investment strategies. He highlights the critical first steps of data curation and feature engineering, noting common errors in data handling. Gurnani contrasts conventional with alternative data and discusses challenges in machine learning transparency. He also shares innovative AI-driven trading strategies and emphasizes the importance of human expertise alongside technology in investment decision-making.
AI Snips
Chapters
Books
Transcript
Episode notes
Five Steps of AI Investing
- Successful AI investing involves five steps: problem framing, data curation, feature engineering, model training, and evaluation. - The greatest edge lies in problem framing, data curation, and feature engineering rather than flashy algorithms alone.
Conventional vs Alternative Data
- Conventional data like financial statements is clean and plentiful; edge comes from interpretation. - Alternative and unstructured data like transcripts and satellite images require heavy human expertise for cleaning and hypothesis mapping.
Handle Alternative Data Carefully
- Use multiple data vendors when working with alternative data like credit card transactions to reduce bias. - Evaluate vendor reliability and panel stability thoroughly to ensure data consistency over time.
