Stock Market Analysis & ML Prediction

A python project for analyzing stocks and building machine learning models to predict price movements for swing trading (2-3 months) and long-term investing (1-3 years).

Project Goals

• Swing Trading: Identify stocks with 10%+ potential gains over 2-3 months
• Long-term Investing: Find fundamentally strong stocks for 1-3 year holds
• Risk Management: Lower-risk approach compared to day trading
• ML-Powered: Use machine learning to identify patterns in historical data

Project Structure

market-analysis/
├── src/                        # Reusable Python modules
│   ├── __init__.py          
│   ├── indicators.py          # Technical indicators (RSI, MACD, etc.)
│   ├── features.py            # Feature engineering for ML
│   ├── data_fetcher.py        # Download stock data
│   ├── models.py              # ML model classes
│   └── signals.py             # Buy/sell signal generation
│
├── notebooks/                  # Jupyter notebooks for analysis
│   ├── 01_swing_trading.ipynb
│   ├── 02_position_trading.ipynb
│   ├── 03_longterm_investing.ipynb
│   └── 04_backtesting.ipynb
│
├── data/                       # Stock data storage
│   ├── raw/                   # Original downloaded data
│   └── processed/             # Cleaned & featured data
│
├── models/                     # Saved ML models
│   └── swing_model_v1.pkl
│
├── results/                    # Charts, reports, backtest results
│   └── analysis_charts/
│
├── requirements.txt           # Python dependencies
├── README.md                  # This file
└── Logbook.md

Core Features

Technical Indicators

• Moving Averages: SMA, EMA (20, 50, 200-day)
• Momentum: RSI, MACD, Stochastic
• Volatility: Bollinger Bands, ATR
• Volume: OBV, Volume trends
• Trend: ADX, Ichimoku Cloud

ML Models

• Random Forest Classifier
• XGBoost
• LightGBM
• Feature importance analysis
• Walk-forward validation

Analysis Types

• Swing Trading (2-3 months): Technical pattern recognition
• Position Trading (6-12 months): Momentum + technical
• Long-term Investing (2-3 years): Fundamentals + growth

Learning Path

Week 1-2: Data & Exploration

1. Fetch data for 5-10 stocks
2. Calculate technical indicators
3. Visualize patterns
4. Understand correlations

Week 3-4: Feature Engineering

1. Create lagged features
2. Add technical indicators
3. Engineer price-based features
4. Analyze feature importance

Week 5-6: First Models

1. Logistic Regression baseline
2. Random Forest classification
3. Compare performances
4. Feature selection

Week 7-8: Backtesting

1. Walk-forward validation
2. Calculate returns
3. Risk metrics (Sharpe ratio)
4. Identify weaknesses

Week 9+: Iterate & Improve

1. Try ensemble methods
2. Add fundamental data
3. Multi-stock screening
4. Portfolio optimization

Dependencies

Core packages:

• yfinance - Download stock data from Yahoo Finance
• pandas - Data manipulation
• numpy - Numerical computing
• scikit-learn - Machine learning models
• xgboost - Gradient boosting
• lightgbm - Fast gradient boosting
• matplotlib - Plotting
• seaborn - Statistical visualization
• plotly - Interactive charts
• jupyter - Interactive notebooks

See requirements.txt for full list with versions.

Data Sources

Primary: Yahoo Finance (via yfinance)

• Free, no API key needed
• historical data going back decades
• OHLCV data (Open, High, Low, Close, Volume)
• Adjusted for splits and dividends
• ~15 minute delay on real-time data

trading Strategy

Swing Trading

• Timeframe: 2-6 months
• Target Return: 10-20%
• Risk Level: Medium
• Analysis: 70% technical, 30% fundamental
• Time Commitment: Few hours per week

Position Trading

• Timeframe: 6-12 months
• Target Return: 20-40%
• Risk Level: Medium-Low
• Analysis: 50% technical, 50% fundamental
• Time Commitment: Monthly review

Long-term Investing

• Timeframe: 2-3+ years
• Target Return: 50-200%+
• Risk Level: Lower (if fundamentally sound)
• Analysis: 80% fundamental, 20% technical
• Time Commitment: Quarterly review

Useful Resources

• Investopedia - Financial education
• QuantInsti Blog - Algorithmic trading
• Machine Learning for Trading - Course
• Yahoo Finance - Market data
• TradingView - Charting platform

Future Improvements

• Add fundamental data integration
• Build portfolio optimizer
• Create stock screener with rankings
• Add sentiment analysis (news, Twitter)
• Implement automated backtesting
• Add risk metrics dashboard
• Create alerting system
• Support multiple data sources
• Add options analysis
• Build web dashboard

Contributing

This is a personal learning project, but suggestions are welcome!

License

MIT License - Feel free to use and modify for your own learning.

Contact

Created by Jasper Valk - Stock Market Analysis Project

---