Order Matching Engine (C++)

• working on it (08/21/25s)

Order Matching Engine (Python Prototype)

• readme copied from prototype/README.md

Overview

This repository contains a prototype limit order book and matching engine implemented in Python. The system is designed for low-latency order handling and performance benchmarking. It provides a reference implementation for message-driven trading infrastructure.

Features

• In-memory limit order book with FIFO matching at each price level
• Support for limit, market, and cancel orders
• JSON-based message stream interface
• Benchmark harness with time.perf_counter for latency measurement
• Separation of parsing, matching, and trade logging for future concurrency extensions

Project Structure

.
├── prototype
│   ├── order_book_matching.py # Core order book data structures and matching logic
│   ├── order_book_skeleton.py # resting order book
│   ├── price_level.py
│   ├── README.md
│   └── throughput_testing.py # benchmarking harness for matching version of orderbook
└── README.md

Performance

Latency benchmarks are collected using random order streams.

Ran on 2021 Macbook Pro M1 Pro

Without JSON parsing

• p50 ≈ 1.0 µs
• p99 ≈ 1.5 µs
• p99.9 ≈ 8–9 µs

With json.loads parsing

• p50 ≈ 2.17 µs
• p99 ≈ 2.67 µs
• p99.9 ≈ 11.63 µs

Results highlight both the efficiency of the order book operations and the overhead introduced by JSON deserialization in Python.

Design

• Order Book:
  Two heaps are maintained: a max-heap for bids and a min-heap for asks. Each price level holds a FIFO queue of orders. Matching executes greedily and supports partial fills.

• Message Processing:
  Incoming messages are JSON objects of type LIMIT, MARKET, or CANCEL. Messages are parsed, applied to the order book, and may generate trades.

• Benchmarking:
  Random order streams are generated and processed. Latencies are recorded per message and aggregated into percentile distributions.

Extensions

• Replace JSON with a binary protocol to reduce parsing overhead
• Add lock-free queues for concurrent parsing, matching, and logging
• Benchmark at higher message rates to evaluate throughput limits
• Implement visualization of order book depth and trade activity