The Alchemist - AI Forex Experimentation & Trading Platform

The Alchemist is an AI Forex Experimentation & Trading Platform that enables iterative development of profitable DRL (Deep Reinforcement Learning) trading models. The platform provides a complete workflow from data collection → feature engineering → model training → backtesting → paper trading → live deployment.

Platform Overview

The Alchemist provides:

• Extensible Data Sources: Auto-discovery system for data providers with feature catalog
• Experiment Builder: Create and configure experiments through the dashboard UI
• Dual Training Modes: Live training (real-time data) or historical backtesting
• Hyperparameter Tuning: Manual configuration or automated Optuna search
• Model Lifecycle: Clear progression from Training → Paper Trading → Live Trading
• Safety First: Kill switch, circuit breakers, and order management built-in
• Performance Tracking: Real-time metrics, equity curves, and trade history
• React Dashboard: Complete experimentation UI for managing experiments

Architecture

┌─────────────────┐
│  React Dashboard│ (Port 3000)
└────────┬────────┘
         │
┌────────▼────────┐
│  FastAPI Service│ (Port 8000)
└────────┬────────┘
         │
┌────────▼────────┐
│ Trading Server  │ (Port 1234)
└────────┬────────┘
         │
    ┌────┴────┐
    │   MT5   │
    └─────────┘

Core Components

1. Data Collection & Features

   • Real-time tick data from MT5
   • Technical indicators (RSI, MACD, Bollinger Bands, etc.)
   • Auto-discovery of features from registered providers
   • Feature catalog with metadata

2. Experiment Management

   • Create experiments with feature selection
   • Configure hyperparameters manually or via Optuna
   • Track experiments with MLflow integration
   • Model registry with promotion workflow

3. Trading Operations

   • Direct connection to MT5 Expert Advisors
   • Order management with idempotency
   • Position reconciliation
   • Real-time trade execution

4. Safety Infrastructure ✅

   • Kill Switch: Multiple trigger mechanisms (file, env, network, signal, API)
   • Circuit Breaker: Automatic halt on excessive losses
   • Order Management System: Complete order lifecycle tracking

5. Performance Tracking ✅

   • Trade logging and P&L calculation
   • Equity curve tracking
   • Performance metrics (Sharpe ratio, win rate, drawdown, etc.)
   • Portfolio-level aggregation

6. MLOps ✅

   • MLflow experiment tracking
   • Model versioning and promotion
   • Paper trading validation
   • Model lifecycle management

Requirements

• Docker and Docker Compose
• Python 3.11
• MetaTrader 5 platform with a valid paper or live account

Quick Start

1. Create a .env file in the project root with the following variables:

   SERVER_IP=0.0.0.0
   SERVER_PORT=1234
   DB_HOST=mariadb
   DB_PORT=3306
   DB_NAME=db_forex
   DB_USER=forex_user
   DB_PASSWORD=forex_password
   MYFXBOOK_EMAIL=your_email@example.com
   MYFXBOOK_PASSWORD=your_password
   URL_MYFXBOOK=https://www.myfxbook.com/

2. Run database migrations (if not already done):

   # Option 1: Use migration runner (recommended)
   python scripts/run-migrations.py
   
   # Option 2: Run scripts manually
   mysql -u root -p db_forex < database/scripts/01_create.sql
   mysql -u root -p db_forex < database/scripts/02_procedures.sql
   mysql -u root -p db_forex < database/scripts/06_safety_infrastructure.sql

3. Start the Docker environment:

   docker compose up -d --build

4. Verify the server is running:

   docker logs server

   You should see: Server socket bind to 0.0.0.0:1234

MT5 Platform Configuration

1. Install MetaTrader 5 from the official website

2. Find your Docker host IP address:

   • On Windows/Mac: Use localhost or 127.0.0.1 if running Docker Desktop
   • On Linux: Use your machine's IP address or localhost
   • If using Docker on a remote server, use that server's IP address

3. Configure Expert Advisors:

   • Navigate to: Tools -> Options -> Expert Advisors
   • Enable "Allow WebRequest for listed URL"
   • Add your endpoint: http://<your-ip>:1234 (if needed for web requests)

4. Load the Expert Advisors:

   • Copy the EAs from src/mql5_code/EAs/ to your MT5 Experts folder
   • In the EA inputs, set:
     • ip: Your Docker host IP (e.g., localhost or your machine's IP)
     • port: 1234 (must match SERVER_PORT in .env)

5. Attach EAs to charts:

   • mt5_tick_streamer.mq5: Attach to any chart for the symbol you want to stream
   • mt5_trading_operation.mq5: Attach to enable trading operations

Documentation

• API Reference - Complete API documentation with examples
• User Guide - Getting started and using the platform
• Developer Guide - Adding data sources and extending features
• Deployment Guide - Setup, configuration, and troubleshooting

Testing

Running Tests

# All tests
pytest tests/

# Unit tests
pytest tests/unit/

# Integration tests
pytest tests/integration/

# API tests
pytest tests/api/

# With coverage
pytest tests/ --cov=src/mt5-python_server/src --cov-report=html

Test Coverage

• Unit tests: Core components and utilities
• Integration tests: Component interactions and workflows
• API tests: All REST endpoints and WebSocket channels
• E2E tests: Complete user workflows
• Performance tests: Benchmarks and optimization validation

Quick Start

1. Clone and Configure

   git clone <repository-url>
   cd alchemist-platform
   cp .env.example .env
   # Edit .env with your settings

2. Start Services

   docker compose up -d

3. Access Dashboard

   • Open http://localhost:3000
   • Create your first experiment!

See Deployment Guide for detailed setup instructions.

License

Proprietary software of Alchemist Capital

Unauthorized use or distribution is prohibited

© 2026, Alchemist Capital - All rights reserved