Dio

High-performance OCaml 5.2 trading engine for Kraken featuring domain-based parallel strategy execution. Each trading asset runs in its own isolated domain with lock-free communication, circuit breaker connection management, and real-time telemetry. Built for high-frequency trading with WebSocket data feeds, asynchronous order execution.

Requirements

• OCaml 5.2.0 (opam)
• Kraken API key/secret
• macOS/Linux/WSL

Quick Start

Install

git clone https://github.com/malciller/dio-domains.git
cd dio-domains
opam install . --deps-only
dune build

Configure

Create .env:

KRAKEN_API_KEY=your_kraken_api_key
KRAKEN_API_SECRET=your_kraken_api_secret
CMC_API_KEY=your_cmc_api_key              # Fear & Greed index (optional fallback to default)

Edit config.json (example):

{
  "logging_level": "info",            // Log verbosity: debug, info, warning, error
  "logging_sections": "",    // Filter logs by section (optional, comma-separated)  
  "trading": [
    {
      "symbol": "BTC/USD",            // Pair to trade
      "qty": "0.0002",                // Base asset quantity per order
      "grid_interval": [0.25, 1.25],  // Min/Max grid spacing (%); resolved once from Fear & Greed
      "sell_mult": "0.999",           // Sell amount multiplier (qty * sell_mult = sell order size)
      "strategy": "Grid"              // Strategy name: "Grid"
    },
    {
      "symbol": "USDG/USD",           // Pair to trade
      "qty": "100.0",                 // Trade size per market making quote
      "min_usd_balance": "500.0",     // Minimum USD balance required to run this strategy. Optional, but required if no max_exposure.
      "max_exposure": "500.0",        // Maximum asset balance allowed before pausing strategy. Optional, but required if no min_usd_balance.
      "strategy": "MM"                // Strategy name: "MM"
    }
  ]
}

Run

./_build/default/bin/main.exe            

Strategies

• GRID: Maintains buy/sell ladders around price with configurable spacing and size.
• MM (Adaptive Market Maker): Dynamically adapts its quoting style based on market fees—uses a greedy quoting approach for no-fee markets and a conservative, profit-guaranteeing strategy where trading fees apply. Always ensures configured profit margins are met by factoring in fees as necessary.
• Fear & Greed: Grid spacing is resolved once at startup using CoinMarketCap Fear & Greed; provide CMC_API_KEY for live values or fallback defaults are used.

Key Features

• Domain-Based Parallelism: Each trading asset runs in its own OCaml domain for true parallel execution without GIL limitations
• Lock-Free Communication: Event-driven architecture with ring buffers for high-throughput data processing
• Circuit Breaker Protection: Automatic connection management with health monitoring and graceful degradation
• Real-Time Telemetry: Comprehensive metrics collection for performance monitoring and debugging
• High-Frequency Trading: Optimized for low-latency execution with microsecond-precision timing
• Fault Tolerance: Supervised domains with automatic restart and exponential backoff

Architecture

• Main Entry Point: Command-line interface with dashboard and headless modes, signal handling for graceful shutdown.
• Domain Spawner: Manages OCaml domains for parallel strategy execution - each trading asset runs in its own isolated domain with supervision and auto-restart capabilities.
• Supervisor: Orchestrates WebSocket connections with circuit breaker patterns, heartbeat monitoring, and connection health management.
• Engine Core:
  • Concurrency: Lock-free event bus and registry for inter-domain communication.
  • Config: Configuration management and validation.
  • Strategies: High-frequency trading algorithms (Grid, Adaptive Market Maker) running in parallel domains.
  • Order Executor: Asynchronous order placement, amendment, and cancellation with duplicate detection.
  • Telemetry: Real-time performance monitoring and metrics collection.
  • Logging: Structured logging with configurable levels and sections.
• Exchange Integration (Kraken):
  • WebSocket Feeds: Real-time ticker, orderbook, balance, and execution data with ring buffer storage.
  • Trading Client: Authenticated order operations with ping/pong heartbeat monitoring.
  • Authentication: Secure token generation and management.

Development

# Build
dune build

# Tests
dune test

# Format / Docs
dune fmt
dune build @doc

# Build and run the project
dune build
./_build/default/bin/main.exe

Logging

• Debug, Info, Warning, Error — timestamped by component.

Contributing

1. Fork
2. Create branch (git checkout -b feature/xyz)
   • Trading strategies: src/engine/strategies/
   • Domain management: src/engine/domain_spawner.ml
   • Connection supervision: src/engine/supervisor/
   • WebSocket feeds: src/exchange/kraken/
   • Dashboard UI: src/ui/
3. Commit (git commit -m "...")
4. Push (git push origin feature/xyz)
5. Open PR

Guidelines: add tests, update docs, keep CI green.

License

MIT — see LICENSE.

Legal: Provided “as is”, without warranty. Trading involves risk.