Access to powerful LLMs like ChatGPT is reshaping roles in the investment profession. This report discusses how to ethically build investment models in the open-source community. It defines alternative data and presents an ESG investing case study.
Report Overview
An artificial intelligence (AI) revolution is here. ChatGPT and other platforms are making powerful large language models (LLMs) accessible to everyone. This is changing how we think about investing and job roles in the investment profession. CFA Institute has long maintained that the future of investing lies in combining AI and human intelligence (HI). The introduction of generative AI (GenAI) may signal a new phase in the AI–HI collaboration.
“Unstructured Data and AI: Fine-Tuning LLMs to Enhance the Investment Process” aims to introduce investment professionals to the concepts shaping a new, more open and more technical investment arena built on the mentality of knowledge sharing and ethical building practices. It seeks to equip readers to leverage alternative data and quantitative techniques for investment modeling, focusing on natural language processing (NLP) and model fine-tuning. It synthesizes the data and tools in the form of a case study on environmental, social, and governance (ESG) investing using alternative data. ESG is an area that is ripe for AI adoption and one for which alternative data can be used to exploit inefficiencies to capture investment returns.
Gain practical understanding of data science and machine learning applications in the investment process.
This paper explores the topic in detail by doing the following:
- Guiding the reader through explanations of alternative and unstructured data, clarifying their differences, and familiarizing the reader with how to ethically start building AI projects with these data in the open-source community.
- Providing necessary background on NLP to begin fine-tuning LLMs and answering questions on what caused such a decisive shift in AI adoption.
- Applying AI concepts in an ESG case study that explores fine-tuning methods to detect material ESG tweets to generate investment returns, highlighting the value of leveraging open-source data and tools to generate new investment ideas.