Newly published research from the CFA Institute Research and Policy Center explores how generative AI tools are creating new opportunities for investing and innovative ways to leverage unstructured, alternative, and open-source data in the investment process.
Platforms like ChatGPT have spurred increased usage and development of powerful large language models (LLMs), quickening the pace of evolving workflows in the industry. Data are being generated at an exponential rate, and the technology powering the algorithms used to parse it is growing just as fast.
The research, Unstructured Data and AI: Fine-Tuning LLMs to Enhance the Investment Process, offers investment professionals a comprehensive framework for ethically building investment models in the open-source community.
The report also features a case study showcasing how unstructured and open-source data can be analysed using open-source NLP tools to enhance investment processes.
The example applies fine-tuning methods to unstructured ESG data (ESG-related tweets from companies in the Russell 1000 Index) to shed light on what types of ESG disclosures resonate most with investors.
Key points
· As data depth and breadth continue to grow and new technologies emerge, investment professionals will need to embrace a more holistic and scientific approach to investing to stay ahead.
· The research offers essential knowledge and information to start building open-source projects – introducing alternative and unstructured data and addressing the importance of ethical considerations while handling these data in the open-source community.
· Natural language processing (NLP) is particularly suited to dealing with alternative and unstructured data. Fine-tuning NLP models on proprietary data can provide demonstrable value.
· The research discusses open-source tools and techniques for building and fine-tuning AI projects, and how to extract insights from open data sources. It also discusses the benefits and challenges of fine-tuning large language models for specific tasks.
· ESG investing seems ripe for AI adoption. The complexities of ESG investing, where data are often unstructured and fragmented, offer an example for exploring fine-tuning methods to detect material ESG information to generate investment returns. A case study illustrates the value of leveraging alternative data and open-source tools to generate new investment ideas.
Join the Webinar: Unstructured Data and AI in Investments
Join the upcoming CFA Institute webinar on Unstructured Data and AI in Investments where panelists will dive into how AI is changing the way investors uncover new insights and hidden value and share the approach to the case study featured in the report.
The webinar will broadcast on 16 May 2024. Register to watch live or on replay.
