Orderbook and Matching Engine

Project Overview

This project implements an efficient orderbook and matching engine in C++. It's designed to scalably handle order placement, cancellation, and modification in a financial trading environment.

Key Features

• Support for multiple order types (GoodTillCancel, FillAndKill, FillOrKill, GoodForDay, Market)
• Efficient matching algorithm for order execution
• Thread-safe operations with mutex locks
• Automatic pruning of expired GoodForDay orders
• Comprehensive order and trade information tracking

Project Structure

The project consists of several key components:

• Orderbook: The main class handling order management and matching
• Order: Represents individual orders with their properties and states
• OrderModify: Handles order modification requests
• Trade: Represents executed trades
• Constants: Defines global constants used in the project
• LevelInfo and OrderbookLevelInfos: Provide information about order book levels

Performance Considerations

• The project leverages efficient data structures for optimal performance:

  • std::unordered_map for quick access to level data and order entries by price and OrderId respectively (O(1) average case lookup).
  • std::map for maintaining sorted bid and ask orders, allowing for efficient order matching and best price determination (O(log n) for insertion, deletion, and finding the best price, with n being the number of unique price levels in the book).
  • std::list for storing orders at each price level, enabling fast insertion and deletion of orders (O(1) when the position is known).

• Modern C++ features are utilized throughout the project:

  • Smart pointers (std::shared_ptr) for automatic memory management of orders.
  • Scoped locks (std::mutex and std::lock_guard) for thread-safe operations.
  • Move semantics and perfect forwarding to minimize unnecessary copying.

• The order book structure is designed for fast lookups and updates:

  • Separate collections for bids and asks allow for quick access to the best prices.
  • Price levels are indexed for rapid order placement and cancellation.

• Efficient matching algorithm:

  • Orders are stored in price-time priority, allowing for fair and quick matching.
  • Partial fills are supported without sacrificing performance.

• Automatic pruning of expired orders:

  • A separate thread handles the removal of GoodForDay orders, preventing the order book from accumulating stale orders without impacting the main execution path.

• Use of std::atomic for thread-safe shutdown flag checking.

These design choices result in a performant orderbook and matching engine capable of handling a large volume of orders and trades without excessive latency.