Technology and Innovation Community

Good evening everyone,

I had the chance to meet some of you at certain CFA UK events where I noticed a certain interest for Python solutions linked with AI/ML and local LLM applications for company analysis. As a result and at the suggestion of certain members, I wanted to share with the community one of my recent open-source Python project on that topic.

In parallel of my studies at Audencia BS and Aston University (MSc FinTech), I am developing a Python script aiming to provide a synthetic overview of the main characteristics of a company and its stock from simply its ticker. The main challenge was to work on a way to process financial data using AI locally, without sending any data to servers on the other side of the globe which is often a contractual clause for private funds when processing their clients' confidential information.

On a side note, one of my motivations for developing this script was the CFA program. I learnt about concepts like skewness, kurtosis and return distributions, but current real-world examples aren't always easy to find online. It helped me gain more confidence when using these notions.

You can find the full code on GitHub: gruquilla/FinAPy: Single-stock analysis using Python and local machine learning/ AI tools (Ollama, LSTM).

The project relies on several libraries, but gets its data using yfinance and pygooglenews mainly. The script includes at the moment:

• A basic synthesis about the company: name, location, sector, industry,...
• A market data analysis, with LSTM machine learning forecast, returns statistics, AI LLM analysis,...
• A ratio-based analysis, computed from the financial statements of the last 3 years with AI LLM interpretation and PESTEL context.

The code uses some AI and ML-oriented libraries, but even if you are not used to Python at all, the readme on its own should allow you to understand how the script operates. It can be a good introduction with real-world examples to deploy your own ML/LLM-Python pipeline.

Great Project Guillaume. Some ideas:

• Try sentiment analysis using LLM as well but omitting ticker information and/or company information to minimize forward-looking bias risk.
• Use standardized financial information from a data vendor API to tackle the formatting issues you mention.
• Ollama AI commentary can be enhanced by adding context e.g. provide news sentiment analysis to Ollama so that it does not only work with financial ratio data.

In any case, it's a great tool. Congrats and thanks for sharing

Good evening everyone,

I had the chance to meet some of you at certain CFA UK events where I noticed a certain interest for Python solutions linked with AI/ML and local LLM applications for company analysis. As a result and at the suggestion of certain members, I wanted to share with the community one of my recent open-source Python project on that topic.

In parallel of my studies at Audencia BS and Aston University (MSc FinTech), I am developing a Python script aiming to provide a synthetic overview of the main characteristics of a company and its stock from simply its ticker. The main challenge was to work on a way to process financial data using AI locally, without sending any data to servers on the other side of the globe which is often a contractual clause for private funds when processing their clients' confidential information.

On a side note, one of my motivations for developing this script was the CFA program. I learnt about concepts like skewness, kurtosis and return distributions, but current real-world examples aren't always easy to find online. It helped me gain more confidence when using these notions.

You can find the full code on GitHub: gruquilla/FinAPy: Single-stock analysis using Python and local machine learning/ AI tools (Ollama, LSTM).

The project relies on several libraries, but gets its data using yfinance and pygooglenews mainly. The script includes at the moment:

• A basic synthesis about the company: name, location, sector, industry,...
• A market data analysis, with LSTM machine learning forecast, returns statistics, AI LLM analysis,...
• A ratio-based analysis, computed from the financial statements of the last 3 years with AI LLM interpretation and PESTEL context.

The code uses some AI and ML-oriented libraries, but even if you are not used to Python at all, the readme on its own should allow you to understand how the script operates. It can be a good introduction with real-world examples to deploy your own ML/LLM-Python pipeline.

Good evening everyone,

I had the chance to meet some of you at certain CFA UK events where I noticed a certain interest for Python solutions linked with AI/ML and local LLM applications for company analysis. As a result and at the suggestion of certain members, I wanted to share with the community one of my recent open-source Python project on that topic.

In parallel of my studies at Audencia BS and Aston University (MSc FinTech), I am developing a Python script aiming to provide a synthetic overview of the main characteristics of a company and its stock from simply its ticker. The main challenge was to work on a way to process financial data using AI locally, without sending any data to servers on the other side of the globe which is often a contractual clause for private funds when processing their clients' confidential information.

On a side note, one of my motivations for developing this script was the CFA program. I learnt about concepts like skewness, kurtosis and return distributions, but current real-world examples aren't always easy to find online. It helped me gain more confidence when using these notions.

You can find the full code on GitHub: gruquilla/FinAPy: Single-stock analysis using Python and local machine learning/ AI tools (Ollama, LSTM).

The project relies on several libraries, but gets its data using yfinance and pygooglenews mainly. The script includes at the moment:

• A basic synthesis about the company: name, location, sector, industry,...
• A market data analysis, with LSTM machine learning forecast, returns statistics, AI LLM analysis,...
• A ratio-based analysis, computed from the financial statements of the last 3 years with AI LLM interpretation and PESTEL context.

The code uses some AI and ML-oriented libraries, but even if you are not used to Python at all, the readme on its own should allow you to understand how the script operates. It can be a good introduction with real-world examples to deploy your own ML/LLM-Python pipeline.