The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence) NLP for Equity Investing with Frank Zhao - #424
4 snips
Nov 2, 2020 
Frank Zhao, Senior Director of Quantamental Research at S&P Global Market Intelligence, delves into the fusion of machine learning and finance. He discusses leveraging unstructured data to gain an edge in equity investing, particularly through natural language processing of earnings call transcripts. Zhao explains foundational techniques like sentiment analysis and the importance of addressing biases in investment models. The conversation also covers building scalable data pipelines and the significance of innovative metrics in enhancing predictive analysis for investors.
AI Snips
Chapters
Transcript
Episode notes
NLP Paper Series and Python Code
- Frank Zhao's team published an NLP paper series focusing on earnings call transcripts.
- They provided 400+ lines of Python code for clients to perform basic NLP analysis.
Two Categories of Predictive Analytics
- Predictive analytics from earnings calls fall into two categories: sentiment and transparency.
- Transparency scores quantify how clearly executives communicate, using metrics like language complexity.
Importance of Economic Intuition and Robustness
- In quantitative research, economic intuition is crucial.
- Robustness is key, ensuring signals work in the future, not just in backtests.