Software of LLMs in Portfolio Administration: Creating Thematic Universe Index

By Ajay Pawar

On this article, we’ll discover the applying of Giant Language Fashions (LLMs) in Finance, masking matters like:

• Quantitative finance
• Portfolio administration
• Sentiment evaluation
• Information evaluation
• OpenAI API and Python

We’ll reveal how LLMs, paired with instruments like OpenAI API and Python, can streamline processes akin to producing thematic portfolios and analysing traits for smarter funding choices.

This weblog is for you if you’re are motivated by:

• Ideation: Studying modern methods to use LLMs in finance and quantitative evaluation.
• Implementation: Gaining sensible steerage on integrating LLMs for duties like information buying and selling methods and sentiment evaluation, thematic investing, or automating monetary analysis.
• Effectivity: Discovering how LLMs simplify advanced, time-intensive workflows in quantitative portfolio administration and monetary evaluation.

Studying Stage: Superior

Conditions

1. You might have the essential concept of utilizing ChatGPT for Buying and selling. You know the way to jot down prompts to create a easy Python code and have been in a position to improve it additional.
2. You might have examine Buying and selling utilizing LLM. You might be conscious of some examples of how Generative AI & Sentiment Evaluation in Finance/Sentiment Evaluation for buying and selling can help merchants to find an edge in risky market conditions.

The way to Create Thematic Universe for Portfolio Methods Utilizing GenAI

What’s Thematic Index investing?

Thematic investing focuses on capturing long-term traits, disruptive applied sciences, or particular sectors that align with an investor’s preferences. One fashionable technique is Thematic Index Investing, the place buyers deal with firms that meet sure standards, akin to sustainability, know-how adoption, or demographic traits.

On this weblog, we reveal create a thematic universe for healthcare firms creating AI options. Doing such analysis manually is time-consuming, resource-intensive, and sometimes susceptible to oversight. By leveraging Generative AI (GenAI), we are able to automate this course of to extract significant insights and create a shortlist of firms aligned with our thematic objectives.

Thematic Focus: Healthcare and AI

Let’s say we wish to put money into healthcare firms which are actively creating AI options. It is a extremely related theme as AI is turning into a crucial part of developments in diagnostics, therapy, and medical analysis.

Downside Assertion

Manually figuring out and analysing firms that match this theme requires combing by way of:

1. Giant datasets just like the S&P 500.
2. Sector-specific data, akin to healthcare firms.
3. Publicly obtainable data like information articles or firm reviews.

This course of shouldn’t be solely resource-intensive but in addition topic to human error.

Our Strategy Utilizing GenAI

By utilizing Generative AI, we automate this course of to:

1. Fetch the S&P 500 knowledge and filter it by sector to isolate healthcare firms.
2. Fetch Information for the Tickers.
3. Analyse obtainable data for every firm, akin to:

• Options: Services or products they supply.
• Know-how: Particular instruments or platforms they’re leveraging.
• R&D: Improvements underneath improvement.

4. Summarise how AI is being utilized by these firms primarily based on publicly obtainable data, akin to information articles.

5. Shortlist the businesses that align with the thematic objective of healthcare and AI.

Outcomes

• Preliminary Universe: The S&P 500 incorporates 62 healthcare firms after filtering for the sector.
• Ultimate Thematic Universe: After making use of the GenAI-powered course of, we recognized 19 firms actively creating or utilising AI options.
• Time Effectivity: What would have taken hours or days of handbook analysis was achieved in minutes utilizing the supplied code.

APIs Used

1. OpenAI API: A paid service that powers the Generative AI evaluation for summarising and classifying firm data.
2. NewsAPI: A free developer model (https://newsapi.org/) with a restrict of 100 requests per day, used to fetch information articles associated to the businesses.

Disclaimer

• This weblog and code are for academic and analysis functions solely.
• We’re not selling or endorsing any firm, model, or product talked about on this weblog.
• A number of the data is LLM-generated and should require cross-validation for accuracy.
• That is not funding recommendation and shouldn’t be construed as a suggestion to put money into any of the businesses talked about.

The way to Create Thematic Universe for Portfolio Methods Utilizing GenAI? – Python pocket book

Login to Obtain

By utilizing this methodology, we showcase how Generative AI can streamline the method of making thematic universes for portfolio methods and determine potential candidates for thematic index investing.

This analysis highlights 19 healthcare companies leveraging AI in areas such as diagnostics, drug development, patient care, surgical precision, research, sustainability, and healthcare tools. The list is based on the last six months of news, constrained by the News API’s data access. While additional companies may fall outside this scope, the curated selection provides a focused starting point.

Generated by LLMs, the analysis may include hallucinations, overlaps, or misinterpretations, necessitating manual validation for accuracy. Despite this, it saves time by narrowing the field for deeper exploration.

Portfolio managers and traders can use this universe alongside quantitative tools to identify market opportunities, optimise timing, and diversify investments. Investors focused on healthcare can overweight high-growth potential companies, while traders can track firms nearing key milestones, such as drug approvals, for active monitoring and strategic market entries.

Summary

Generative AI is transforming investment research by offering innovative tools to uncover opportunities and trends with unparalleled speed and precision.

In this blog, we explored how Generative AI can assist in creating a thematic universe for healthcare companies actively developing AI solutions. Our approach involved filtering the S&P 500 healthcare companies, gathering relevant news, analysing it with large language models (LLMs), and extracting critical information in a structured format. By incorporating an AI-specific keyword filter, we identified companies working on innovative technologies and solutions.

This process generated a comprehensive database outlining each company’s solutions, technologies, and R&D efforts. Such a database can be leveraged to create or refine themes. For instance, companies focused on diabetes/cancer solutions can be easily identified from the existing approach. Traditionally, conducting such an extensive study across a large universe could take weeks. However, our analysis achieved a focused output within minutes, significantly reducing time and labor. Nevertheless, manual validation remains essential to ensure accuracy, as the quality of insights heavily relies on available data and LLM models.

For investors and traders, this approach adds a new dimension to decision-making. Healthcare-focused investors can strategically overweight specific categories of companies with strong growth potential, while traders can actively monitor firms nearing key milestones, such as drug approvals. By combining this thematic analysis with quantitative tools, market participants can optimise timing, diversify investments, and uncover actionable opportunities more effectively.

Takeaways:

In this blog we covered:

• Ideation: Generating new ideas for leveraging LLMs in portfolio management, sentiment analysis, and thematic investing.
• Implementation: Practical steps to apply tools like the OpenAI API and Python for automating research workflows.
• Use Cases: Real-world examples of AI integration into finance by constructing thematic universes.
• Tools & Techniques: Insights into combining LLMs with quantitative tools to uncover and act on market opportunities effectively.

Conclusion

Generative AI offers a scalable, efficient blueprint for constructing thematic universes tailored to investor goals and market opportunities. As this technology evolves, its role in refining investment strategies will become even more vital. By balancing AI-driven insights with human expertise, market participants can make smarter, more informed, and efficient decisions, capitalising on trends in the healthcare industry and beyond.

Relevant Advanced Courses on Quantra

1. For Traders:

Utilise Large Language Models (LLMs) to build sentiment-driven trading strategies. This course covers LLM basics, prompt engineering for actionable insights, and practical use in trading. Learn to extract sentiment scores from event transcripts like FED meetings or earnings calls, and develop different strategies around it. Analyse your strategy’s performance rigorously, leveraging LLM capabilities for informed trading decisions.

Course link: Trading Using LLM: Concepts and Strategies

2. For Portfolio Managers:

Are you looking to use AI in trading to figure out how much to invest in different stocks? This course has got the answers, thanks to LSTM networks. This course covers fundamental portfolio management with mean-variance optimisation and practical application of AI algorithms. Master walk-forward optimisation, hyperparameter tuning, and real-world portfolio management. Gain hands-on experience with live trading templates and capstone projects.

Course link: AI for Portfolio Management: LSTM Networks

File in the download

• Steps to Create Thematic Universe for Portfolio Strategies Using GenAI – Python notebook

Login to Download

About the Author
Ajay Pawar is a Quantitative Researcher and Analyst at Quantinsti, specializing in Computational Finance, Algorithmic Trading, Data Science, and Portfolio Management. In his prior roles at CRISIL, Axis Securities, BITA and HTTS, he has developed advanced quantitative tools, stock selection models, and algorithmic solutions. Ajay holds an M.Sc. in Financial Data Intelligence from Rennes School of Business, a B.Sc. Economics From Symbiosis School of Economics and is an EPAT (Executive Programm in Algorithmic Trading) Alumnus, He has several other certifications in Business Analytics and Data Science as well. His strong academic foundation and technical expertise make him a standout professional in quantitative finance.