NBA GM Simulator with Multi-Agent Trading

A Next.js and Flask application that allows users to chat with NBA team GMs and simulate a multi-agent trading system where AI-powered GMs autonomously negotiate with each other.

Project Overview

This project combines:

1. Chat Interface: Talk to AI-powered NBA team GMs for insights and strategy
2. Trading System: Simulate being a GM and make trades with AI-powered team GMs
3. Agent-to-Agent Communication: AI GMs negotiate trades with each other based on team needs and player valuations

Project Structure

• /frontend: Next.js application with UI components
• /mcp-client: Python client for MCP and trading system backend
• /mcp-server: MCP server for NBA data
• /nba-mcp-server: MCP server with NBA API integration

Features

Chat Mode

• Select any NBA team and chat with their GM
• Ask about team strategy, players, and insights
• Leverages Claude to generate realistic GM responses

GM Mode

• Take control of any NBA team as the GM
• View team roster, salary information, and league activity
• Propose trades to other teams
• Receive counter-offers
• Watch AI GMs make trades with each other
• League simulation that progresses the state of teams

Trading System Architecture

The trading system is built on a multi-agent architecture where each team has an AI agent that:

1. Evaluates Players: Considers stats, contract, age, and position
2. Analyzes Team Needs: Identifies positional needs and roster holes
3. Makes Strategic Decisions: Decides whether to accept, reject, or counter trade offers
4. Proposes Trades: Initiates trades with other teams based on needs
5. Validates Trades: Ensures trades comply with simplified NBA rules

Getting Started

Frontend Setup

1. Install dependencies:

cd frontend
npm install

2. Run the development server:

npm run dev

3. Open http://localhost:3000 in your browser.

Backend Setup

1. Install dependencies:

cd mcp-client
pip install -r requirements.txt

2. Run the Flask server:

python flask_server.py ../nba-mcp-server/nba_server.py

How It Works

Backend Components

• GMAgent: AI agent that represents each team's GM
• LeagueState: Central data model tracking teams, players and trades
• Flask Server: API endpoints for chat and trade functionality

Frontend Components

• TeamDashboard: UI for viewing team roster and league activity
• TradeModal: Interface for proposing and responding to trades
• LeagueActivity: Feed showing recent trades and negotiations

Testing

To test the trading system functionality:

cd mcp-client
python test_trades.py

This will run through test scenarios for:

• Trade proposal acceptance
• Trade proposal rejection
• Counter offers
• Agent-to-agent trading

Future Enhancements

• More sophisticated player valuation algorithms
• Draft pick trading
• Salary cap exceptions
• Three-team trades
• Season simulation
• Win-loss record tracking

Technologies Used

• Next.js & React: Frontend UI
• TailwindCSS: Styling
• Flask: Backend API
• Claude AI: GM conversations and trade evaluations
• MCP Protocol: Multi-agent communication
• NBA API: Team and player data

Credits

• NBA data provided by NBA API
• Original MCP server code by obinopaul on GitHub
• Created for the MCP Hackathon