This project provides an AI-powered stock screener that transforms any Large Language Model (LLM) into a financial analyst. It addresses the need for sophisticated, natural language-driven financial data querying by downloading historical stock data and enabling users to ask complex questions in plain English. The target audience includes engineers, researchers, and power users seeking to leverage AI for financial analysis without extensive manual data processing or complex SQL queries.

How It Works

The system downloads quarterly and annual financial data for stocks from EOD Historical Data. This data is then persisted in both MongoDB for local access and Google BigQuery for scalable querying. An AI-driven natural language interface allows users to query this stored financial data. Queries are translated into SQL, executed against BigQuery, and results are returned, offering insights into stock performance and financial health.

Quick Start & Requirements

• Primary install / run command: git clone the repository, cd into it, and run npm install. Use npm run build followed by node dist/index.js for production, or ts-node index.ts / ts-node upload.ts / ts-node chat.ts for development.
• Non-default prerequisites: Node.js (v18+), npm, MongoDB, Google Cloud Platform (GCP) account with BigQuery enabled, Requesty API key, EOD Historical Data API key. Optional: Ollama for local LLM capabilities.
• Configuration: A .env file is required in the root directory containing MongoDB connection strings, API keys, and GCP service account credentials.
• Links:
  • Ollama: ollama.com/download
  • Requesty: Referral link provided in documentation.
  • EOD Historical Data: Referral link provided in documentation.

Highlighted Details

• AI-powered natural language interface for querying financial data stored in BigQuery.
• Data ingestion pipeline from EOD Historical Data, supporting quarterly and annual financial statements.
• Dual data storage in MongoDB and Google BigQuery.
• Supports both cloud-based (Requesty) and local (Ollama) LLMs for query processing.

Maintenance & Community

The provided documentation does not detail notable contributors, sponsorships, partnerships, or community channels such as Discord or Slack.

Licensing & Compatibility

• License type: MIT License.
• Compatibility: The MIT license generally permits commercial use and linking with closed-source projects, though users should verify specific dependencies.

Limitations & Caveats

Error handling is described as basic and may require enhancement for production environments. Users must manage API rate limits for EOD Historical Data and be aware of potential BigQuery costs associated with data storage and querying. Robust data validation beyond numeric field filtering may be necessary. Security of API keys and credentials is the user's responsibility.