QuantMind is an intelligent knowledge extraction and retrieval framework designed for quantitative finance. It addresses the challenge of information overload by transforming unstructured financial content—papers, news, blogs, and SEC filings—into a queryable semantic knowledge graph. This enables institutional investors, hedge funds, and research teams to conduct AI-powered research at scale, accelerating the discovery of alpha-generating insights in factor strategies, risk models, and market trends.

How It Works

QuantMind employs a decoupled, two-stage architecture for robustness and flexibility. Stage 1, Knowledge Extraction, ingests data from diverse sources (e.g., arXiv, news feeds, financial blogs), parses text, tables, and figures, tags content by research area, and orchestrates an extraction pipeline into a structured knowledge base. Stage 2, Intelligent Retrieval, converts this structured knowledge into high-dimensional embeddings for semantic search. It supports advanced retrieval patterns, including DeepResearch for complex multi-hop reasoning across documents and Retrieval-Augmented Generation (RAG) for natural language Q&A.

Quick Start & Requirements

• Prerequisites: Python 3.8+.
• Installation: Requires uv for package management. Clone the repository (git clone https://github.com/LLMQuant/quant-mind.git), create and activate a virtual environment (uv venv, source .venv/bin/activate or .venv\Scripts\activate), and install dependencies with uv pip install -e ..
• Documentation: Usage examples are provided in the README for single paper processing, batch runs, and resolving free-form intents.

Highlighted Details

• Accepted at the NeurIPS 2025 GenAI in Finance Workshop.
• Transforms unstructured financial documents into a semantic knowledge graph for AI-driven exploration.
• Offers multiple retrieval scenarios: DeepResearch (multi-hop reasoning), RAG (Q&A), and Data MCP (structured data access).

Maintenance & Community

The project is actively under development, with a roadmap indicating ongoing work, including migration to the OpenAI Agents SDK (PRs #71, #6, #7). A detailed "Contributing.md" outlines setup and standards for contributors. No explicit community channels like Discord or Slack are listed.

Licensing & Compatibility

QuantMind is released under the MIT License, which generally permits commercial use and integration into closed-source projects without significant copyleft restrictions.

Limitations & Caveats

The project is undergoing a significant migration to the OpenAI Agents SDK, with core memory and store layers scheduled for future releases. Some advanced features described in the "Vision" section represent aspirational goals and are not yet implemented. The FilesystemMemory component is noted as a future conceptual example.