AlphaForgeX v3 — Automated Quant Intelligence Pipeline

AlphaForgeX is a sophisticated quantitative trading system that automates the entire workflow from data collection to portfolio management. It combines machine learning, feature engineering, and walk-forward backtesting to generate trading signals and portfolio weights.

🎯 Features

• Automated Data Pipeline: Auto-download and cache stock data for 50+ companies
• Feature Engineering: 14 scale-invariant technical signals (momentum, volatility, trend)
• ML-Driven Signals: Cross-sectional ranking system (LONG/HOLD/SHORT)
• Walk-Forward Backtesting: Expanding window validation with transaction costs
• Portfolio Optimization: Inverse-volatility weighting with confidence scoring
• Interactive Dashboard: Bloomberg-style UI with comprehensive analytics
• Real-Time Analysis: Live signal generation and performance tracking
• Signal Diagnostics: Quality analysis and predictive power assessment

📋 Prerequisites

• Python 3.8+
• pip (Python package manager)
• Git (for version control)
• Windows/macOS/Linux

🚀 Quick Start

1. Installation

Clone the repository:

git clone https://github.com/aadityat23/ALPHAFORGEX.git
cd ALPHAFORGEX

Install dependencies:

pip install -r requirements.txt

2. Run the Main Pipeline

Default mode (analyzes 50 stocks):

python main.py

Custom universe:

python main.py --universe SP50

Custom tickers:

python main.py --tickers AAPL,MSFT,GOOGL

3. Launch Interactive Dashboard

streamlit run dashboard/app.py

The dashboard opens at http://localhost:8501 with:

• Overview: Portfolio summary and key metrics
• Performance: Returns, Sharpe ratio, drawdown analysis
• Risk Engine: Volatility, correlation, VaR metrics
• Alpha Signals: Real-time trading signals
• Screener: Stock filtering and ranking
• Forecasting: Price prediction and probability analysis
• Reports: Detailed performance and trade analysis
• Balance Sheet: Financial metrics and fundamentals

📊 Pipeline Overview

1. Load Data
   ↓
2. Engineer Features (14 signals)
   ↓
3. Create Targets (Cross-sectional ranking)
   ↓
4. Walk-Forward Backtest (Expanding window)
   ↓
5. Signal Diagnostics (Quality analysis)
   ↓
6. Train Final Model (All data)
   ↓
7. Generate Live Signals
   ↓
8. Portfolio Optimization (Inverse-vol weights)
   ↓
9. Performance Report (Sharpe, drawdown, win rate)

📁 Project Structure

AlphaForgeX/
├── main.py                          # Main entry point
├── requirements.txt                 # Python dependencies
├── README.md                        # This file
├── pipeline/                        # Core pipeline modules
│   ├── __init__.py
│   ├── data_loader.py              # Data loading & caching
│   ├── feature_engine.py           # Feature engineering (14 signals)
│   ├── target_engine.py            # Target/label creation
│   ├── model_engine.py             # ML model training & prediction
│   ├── backtest_engine.py          # Walk-forward backtesting
│   ├── portfolio_engine.py         # Portfolio optimization
│   ├── diagnostics.py              # Signal quality analysis
│   └── universe.py                 # Stock universe definitions
├── dashboard/                       # Streamlit interactive UI
│   └── app.py                      # Dashboard application
├── data/                           # Data directory
│   ├── features/                   # Pre-computed features (per-ticker CSVs)
│   ├── AAPL.csv, AMZN.csv, ...     # Raw price data
│   └── returns.csv                 # Returns data
├── backtesting/                    # Backtesting utilities
│   └── backtest.py                # Backtest execution
├── data_pipeline/                  # Data fetching
│   └── fetch_data.py              # Data acquisition script
├── feature_engine/                 # Feature computation
│   └── features.py                # Feature definitions
├── models/                         # Model training
│   └── train_model.py             # Model training script
├── portfolio_engine/               # Portfolio management
│   └── optimizer.py               # Portfolio optimizer
├── output/                        # Output results
│   ├── backtest_trades.csv       # Backtest trades
│   ├── current_signals.csv       # Live signals
│   ├── performance_metrics.csv   # Metrics
│   └── portfolio_weights.csv     # Portfolio weights
└── utils/                         # Utilities

🎨 Key Components

Data Pipeline

• Auto-Download: Fetches data from yfinance
• Caching: Local CSV storage for fast access
• Preprocessing: Alignment and normalization

Feature Engineering

14 scale-invariant signals including:

• Momentum indicators
• Volatility metrics
• Trend-following signals
• Mean reversion indicators

ML Model

• Algorithm: Random Forest / Linear Regression
• Training: Walk-forward expanding window
• Output: Probability scores for each signal

Backtesting

• Expanding Window: Realistic forward-testing
• Transaction Costs: Slippage and commission modeling
• Performance Metrics: Sharpe ratio, max drawdown, win rate

Portfolio Optimization

• Weighting: Inverse-volatility × confidence scores
• Diversification: Cross-sectional position sizing
• Risk Management: Volatility targeting

📈 Output Files

After running python main.py, check the output/ folder:

• backtest_trades.csv: All trades from walk-forward backtest
• current_signals.csv: Latest trading signals for all stocks
• performance_metrics.csv: Strategy performance summary
• portfolio_weights.csv: Recommended portfolio weights

⚙️ Configuration

Edit main.py to customize:

FEATURE_DIR       = "data/features/"      # Feature data location
FORWARD_DAYS      = 5                     # Prediction horizon (days)
LONG_PCT          = 0.25                  # Top % → LONG (25% = ~12 stocks)
SHORT_PCT         = 0.25                  # Bottom % → SHORT

🔧 Dependencies

• pandas: Data manipulation
• numpy: Numerical computing
• yfinance: Stock data fetching
• scikit-learn: Machine learning
• matplotlib: Plotting
• streamlit: Interactive dashboard

📊 Typical Workflow

1. Development: Modify features in pipeline/feature_engine.py
2. Testing: Run python main.py to test pipeline
3. Analysis: Launch dashboard to visualize results
4. Refinement: Adjust model parameters and re-run
5. Deployment: Use generated signals for trading

🐛 Troubleshooting

Issue: Data download fails

• Solution: Check internet connection, ensure yfinance is installed

Issue: Dashboard won't start

• Solution: Run pip install streamlit and ensure port 8501 is available

Issue: Features not computing

• Solution: Verify data/features/ directory exists and contains CSV files

Issue: Model training takes too long

• Solution: Reduce stock universe or adjust backtest window in main.py

📝 Example Output

╔══════════════════════════════════════════════════════════════════╗
║                     AlphaForgeX Pipeline Complete                ║
├──────────────────────────────────────────────────────────────────┤
║ Stocks Analyzed:        50
║ Features Engineered:    14
║ Walk-Forward Periods:   12
║ Model Accuracy:         58.3%
║ Sharpe Ratio:           1.45
║ Max Drawdown:           -12.3%
║ Win Rate:               55.2%
║ Recommended Actions:    LONG: 12, HOLD: 26, SHORT: 12
╚══════════════════════════════════════════════════════════════════╝

🚀 Deployment

To use live trading signals:

1. Run python main.py regularly (daily/weekly)
2. Check output/current_signals.csv for recommendations
3. Use output/portfolio_weights.csv for position sizing
4. Execute trades through your broker

📚 Learning Resources

• Quantitative Trading: Understand ML in finance
• Feature Engineering: Domain-specific signal creation
• Walk-Forward Analysis: Realistic backtesting methodology
• Portfolio Optimization: Risk-adjusted weighting

📄 License

This project is provided as-is for educational and research purposes.

👤 Author

Created as an advanced quantitative trading system.

🤝 Contributing

Contributions welcome! Areas for enhancement:

• Additional feature indicators
• Alternative ML algorithms
• Advanced portfolio constraints
• Real-time execution integration
• Risk factor analysis

---

Disclaimer: This system is for educational purposes. Past performance does not guarantee future results. Use at your own risk and always perform thorough backtesting before live trading.