QuantConnect Research System

An AI-powered quantitative finance research platform built on the Open Agent System pattern. This system leverages QuantConnect's Lean engine patterns to analyze trading strategies, track live traders, and synthesize optimal portfolios.

Features

• Algorithm Analysis: Scrape and analyze QuantConnect Lean algorithms
• Alpha Detection: Identify genuine alpha signals with statistical rigor
• Live Trader Tracking: Monitor on-chain traders across DeFi and prediction markets
• Portfolio Optimization: Combine multiple strategies for optimal risk-adjusted returns
• Professional Metrics: Sharpe, Sortino, PSR, Alpha/Beta, and 20+ other metrics

Quick Start

# Clone the repository
git clone https://github.com/yourusername/quantconnect.git
cd quantconnect

# Install dependencies
pip install -r requirements.txt

# Use with Claude Code
claude

Available Skills (Slash Commands)

Command	Description
/quant-scrape	Fetch and analyze Lean algorithms
/alpha-analyze	Evaluate strategy for alpha signals
/strategy-combine	Optimize multi-strategy portfolios
/backtest-eval	Evaluate backtest results
/track-trader	Monitor live traders

Example Usage

Analyze a Lean Algorithm

/quant-scrape MACDTrendAlgorithm

Evaluate Strategy Alpha

/alpha-analyze data/backtest_results.csv --benchmark SPY

Track a DeFi Trader

/track-trader 0x1234...5678 --platform uniswap --alerts

Combine Strategies

/strategy-combine momentum mean-reversion --max-drawdown 20 --method max_sharpe

Architecture

┌─────────────────────────────────────────────────────────────────┐
│                        CLAUDE CODE                              │
│  ┌─────────────────────────────────────────────────────────────┤
│  │ Skills: /quant-scrape /alpha-analyze /strategy-combine      │
│  ├─────────────────────────────────────────────────────────────┤
│  │ Agents: Lean Scraper | Alpha Analyzer | Strategy Combiner   │
│  │         Trade Analytics | Live Tracker | Quant Researcher   │
│  ├─────────────────────────────────────────────────────────────┤
│  │ Tools: fetch_lean_algorithm | calculate_metrics | parse_algo│
│  └─────────────────────────────────────────────────────────────┘
└─────────────────────────────────────────────────────────────────┘
                              │
          ┌───────────────────┼───────────────────┐
          ▼                   ▼                   ▼
    ┌───────────┐       ┌───────────┐       ┌───────────┐
    │QuantConn  │       │ On-Chain  │       │ Prediction│
    │   Lean    │       │   Data    │       │  Markets  │
    └───────────┘       └───────────┘       └───────────┘

Project Structure

quantconnect/
├── open-agents/               # Open Agent System
│   ├── INSTRUCTIONS.md        # System configuration
│   ├── agents/                # Agent definitions
│   ├── tools/                 # Python tools
│   └── outputs/               # Generated outputs
├── .claude/skills/            # Claude Code skills
├── data/                      # Data storage
│   ├── raw/                   # Scraped data
│   ├── processed/             # Analysis results
│   └── cache/                 # API cache
├── docs/                      # Documentation
└── src/                       # Additional code

Metrics Computed

Performance

• Total Return / CAGR
• Volatility (annualized)
• Maximum Drawdown

Risk-Adjusted

• Sharpe Ratio
• Sortino Ratio
• Calmar Ratio
• Probabilistic Sharpe Ratio (PSR)

Factor Analysis

• Alpha (vs benchmark)
• Beta (market sensitivity)
• Information Ratio
• Treynor Ratio

Trade Statistics

• Win Rate
• Profit Factor
• Expectancy
• Kelly Fraction

Data Sources

Source	Type	Purpose
QuantConnect Lean	GitHub	Algorithm examples
The Graph	Subgraph	DeFi trading data
Polymarket	GraphQL	Prediction market positions
Dune Analytics	SQL	Historical on-chain data

Requirements

• Python 3.11+
• Claude Code CLI
• Git

Dependencies

pandas>=2.0
numpy>=1.24
pyyaml>=6.0
requests>=2.31

Contributing

Contributions welcome! See ARCHITECTURE.md for system design details.

License

MIT License - see LICENSE for details.

Acknowledgments

• QuantConnect for the Lean Engine
• Open Agent System for the agent pattern