AI Trading System

A modular, production-oriented AI-powered quantitative trading system written in Python.

It combines historical market data, rich technical feature engineering, machine-learning signal generation, volatility-targeted position sizing, stop-loss / take-profit rules, and a multi-asset backtester with realistic transaction costs.

Disclaimer: This is research / educational software only.
It is not financial advice. Past performance does not guarantee future results.
Always paper-trade and fully understand the risks before using real capital.

---

Fresh Clone – Exact Commands Buyers Run

git clone <repo-url> ai_trading_system
cd ai_trading_system

python -m venv .venv
source .venv/bin/activate          # Windows: .venv\Scripts\activate

pip install -r requirements.txt
pip install -r requirements-dev.txt

# THIS IS THE CRITICAL COMMAND – must succeed
pytest tests/ -v --cov=src --cov-report=term-missing

# Alternative
make test

Run tests in a completely isolated container

docker compose run --rm test
# or
make docker-test

Run the trading pipeline

python main.py
# or
make run

---

What the Test Suite Covers

File	What it tests
tests/test_smoke.py	Imports, config loading, basic objects
tests/test_data_fetcher.py	Data fetching + caching (mocked)
tests/test_features.py	Technical feature engineering
tests/test_predictor.py	ML model fit / predict / save / load
tests/test_signal.py	Signal generation from probabilities
tests/test_risk.py	Position sizing, stops, drawdown checks
tests/test_backtest.py	Full backtest engine + metrics

All tests run without network access or external credentials.

---

Project Structure

ai_trading_system/
├── .github/workflows/ci.yml      # Lint + type + test on every push
├── config/config.yaml
├── src/
│   ├── data/
│   ├── features/
│   ├── models/
│   ├── strategies/
│   ├── risk/
│   ├── backtest/
│   ├── portfolio/
│   └── utils/
├── tests/                        # Full pytest suite
├── main.py
├── Dockerfile                    # Multi-stage (production + test)
├── docker-compose.yml
├── Makefile
├── pytest.ini
├── requirements.txt              # Pinned runtime
├── requirements-dev.txt
├── requirements-lock.txt
├── pyproject.toml
├── .env.example
├── CONTRIBUTING.md
└── README.md

---

Development Commands

make help
make install-dev
make test
make lint
make format
make ci                 # full local CI
make docker-test        # tests inside Docker
make docker-run         # pipeline inside Docker

---

Configuration

Edit config/config.yaml for symbols, model type, risk limits, etc.

Copy .env.example → .env only if you later add live broker keys (never commit secrets).

---

CI

GitHub Actions (.github/workflows/ci.yml) runs on every push and pull request:

• Multi-Python matrix (3.10 / 3.11 / 3.12)
• Ruff lint + format check
• Full pytest suite with coverage
• Bandit + Safety security scans

---

Extending

• New model → implement fit / predict_proba / save / load in src/models/
• Change prediction target → edit _create_target in SignalPredictor
• Live trading → replace backtester with a broker connector

---

License

MIT