AutoQuant
AutoQuant is an AI-driven hedge fund that executes swift trades using the latest financial news for maximum investment yield, powered by a team of collaborating AI agents built on AutoGen.
- Date
- November 2023
- Role
- AI Engineer
- Tags
- AI Agents, AG2, Multi-Agent Systems, LLM, Algorithmic Trading, FinTech, Python, Hackathon
Overview
AutoQuant is an AI-driven hedge fund experiment that executes swift trades using the latest financial news for maximum investment yield. By retrieving a stream of financial news, AutoQuant performs financial analysis with a team of AI agents to provide recommendations and execute trades based on the news provided. It was built during the AutoGen 24 Hours Hackathon — check out the full submission on lablab.ai.
Presentation
The Problem
Algorithmic trading lacks precision when taking into account financial news and market sentiment with regards to investor bias, creating false cases in the efficient market theory. With the invention of LLMs, AI can properly interpret the news and determine with strong probability how to act on them.
How it's Made
AutoQuant is built around AG2 (formerly AutoGen), a multi-agent framework which orchestrates a team of specialized LLM-powered agents that converse and collaborate to reach a decision. A stream of financial news is fed into the system, where the agents interpret the reporting, weigh the sentiment against market context, produce trade recommendations, and then execute trades based on their collective analysis.
Market Opportunity
- TAM: 14 million active investors trading per year
- SAM: 10% of active investor trading
- SOM: 10% of SAM at $50/year ≈ $70,000,000 / year
Technologies Used
- AG2 (formerly AutoGen) for multi-agent orchestration
- Large Language Models for news interpretation and sentiment analysis
- Financial news stream for real-time market signals
- Python
Project Credits
Built by the YCM hackathon team for the AutoGen 24 Hours Hackathon hosted by lablab.ai.