VelociTrade

A compact trading lab that demonstrates the core building blocks of low-latency systems:

• Market Data (UDP): exchange publishes top-of-book snapshots at ~10 kHz
• Order Path (TCP + Framing): length-prefixed messages, heartbeats, reconnects
• Order Book: minimal L2 structure with atomic snapshot updates
• Strategy: queue imbalance → join best bid/ask
• Risk Controls: price band, notional cap, per-second rate limit
• Latency Measurement: per-order round-trip times in nanoseconds, logged to a binary file and visualized with Python

🔧 Quick Start

✅ Build

mkdir -p build && cd build
cmake .. && cmake --build . -j

🚀 Run

In two terminals:

Terminal A (exchange simulator):

./ex_sim ../config.yaml

Terminal B (client/strategy):

./uclient ../config.yaml 30000

This will simulate sending 30,000 orders at high speed.

📈 Plot Latency

python3 ../tools/plot_latency.py ../latency.bin

This generates a histogram and saves it as latency.png.

📁 Project Structure

.
├── CMakeLists.txt
├── config.yaml
├── include/
│   ├── core/        # timing, logging, framing, config utils
│   ├── proto/       # packed wire structs
│   ├── book/        # minimal L2 order book
│   └── net/         # UDP networking helpers
├── src/
│   ├── sim/ex_sim.cpp      # exchange simulator (market data + TCP order handling)
│   └── client/uclient.cpp  # trading client: book → strategy → risk → orders
└── tools/
    └── plot_latency.py     # binary latency log → matplotlib histogram

⚙️ Configuration (config.yaml)

md_ip: 127.0.0.1
md_port: 5001
ord_ip: 127.0.0.1
ord_port: 9001
levels: 8
risk_band_ticks: 20
risk_notional_cap: 1000000000
risk_max_rate: 2000
heartbeat_ms: 200
reconnect_ms: 500

📦 Requirements

• C++20 compiler (GCC ≥ 11 recommended)
• CMake ≥ 3.16
• Python 3 with matplotlib (for latency plots)

Ubuntu / WSL Install

sudo apt update
sudo apt install build-essential cmake python3-matplotlib -y