Trading Primitives

Low-level components for trading systems in Python.

Components

• Real-time PnL engine: Position tracking + realized/unrealized attribution
• Limit order book: Price-time priority matching
• Position tracker: Net positions + average cost per account/symbol
• Trading engine: Order → match → fill → position → PnL workflow
• Persistence: SQLAlchemy models + repository pattern
• API layer: FastAPI + Pydantic

Production patterns for finance engineering.

Quickstart

docker-compose up -d
uvicorn api.main:app --reload
curl -X POST "http://localhost:8000/trades" \
  -H "Content-Type: application/json" \
  -d '{"symbol": "AAPL", "side": "BUY", "quantity": 100, "price": 150}'

Architecture

graph TD
    A[FastAPI trades POST] --> B[TradingEngine]
    B --> C[OrderBook match]
    C --> D[Fills]
    D --> E[PositionTracker]
    D --> F[PnLEngine]
    E --> G[Postgres Snapshots]
    F --> H[Redis Hot State]
    
    I[Market Data WS] --> J[PnLEngine on price]
    J --> H
    
    K[GET pnl] --> L[In-memory Query]

Loading

Usage Examples

# End-to-end workflow
from core.trading_engine import SimpleTradingEngine

engine = SimpleTradingEngine("AAPL")
order = Order(order_id=1, side="BUY", quantity=100, price=150)
fills = engine.submit_order("PM1", order)
print(f"Fills: {len(fills)}")
print(f"Total PnL: {engine.pnl_engine.get_total_pnl()}")