ML-3m-trader: XAUUSDc 3-Minute Timeframe ML Trading System

An end-to-end proprietary machine learning pipeline for trading XAUUSDc (Gold) on the 3-minute timeframe. This system utilizes a high-performance architecture bridging Python for data processing and model orchestration with Rust for high-frequency execution components.

Project Overview

The ML-3m-trader repository provides a robust framework for automated trading, featuring a hybrid implementation designed for speed and reliability. The core logic involves sophisticated feature engineering and a classification-based approach to market decision-making. Read paper at SSRN for brigde system framework.

Confidentiality Notice: The specific machine learning algorithms and proprietary trading strategies utilized in this system are currently private. The documentation focuses on infrastructure and architectural workflows.

Key Features

• Multi-Language Architecture: Seamless integration between Python processing and Rust execution.
• Data Acquisition: Automated 3-minute OHLCV data fetching.
• Proprietary Labeling: Advanced market state labeling engine with built-in risk-reward and spread filtering.
• Vectorized Backtesting: High-speed, realistic execution modeling accounting for slippage and spread.
• Comprehensive Metrics: Detailed performance analysis including Sharpe, Sortino, and Profit Factor.

Output Preview

The following visualizations illustrate the system's internal processing and performance evaluation.

Feature Processing Workflow

The diagram above details the data transformation pipeline from raw market indicators to model-ready features.

Performance Metrics

The image above showcases the standardized backtesting report generated after a full simulation run.

System Architecture

graph TD
    A[MetaTrader 5] -->|OHLCV Data| B(Data Fetcher)
    B --> C(Feature Engineering)
    C --> D(Labeling Engine)
    D --> E(ML Pipeline)
    E --> F{Backtesting}
    F -->|Performance| G(Metrics Report)
    F -->|Execution| H[Live Trading Interface]
    
    subgraph "Hybrid Processing"
    C
    D
    E
    end

Project Structure

Detailed overview of all project components:

ML-3m-trader/
├── python_version/          
│   ├── main.py
│   ├── config.py
│   ├── data_fetcher.py
│   ├── diag_mt5.py
│   ├── features.py
│   ├── labeler.py
│   ├── model.py
│   ├── backtester.py
│   ├── metrics.py
│   └── README.md
├── rust_ml_trader/         
│   ├── src/
│   │   ├── main.rs
│   │   ├── backtester.rs
│   │   ├── config.rs
│   │   ├── data_fetcher.rs
│   │   ├── features.rs
│   │   ├── labeler.rs
│   │   ├── metrics.rs
│   │   ├── model.rs
│   │   └── types.rs
│   ├── .gitignore
│   ├── Cargo.toml
│   ├── GUIDE.md
│   ├── LICENSE
│   └── README.md
├── SUM3API (local)/        
│   ├── MQL5/
│   │   ├── Experts/
│   │   │   └── ZmqPublisher.mq5
│   │   ├── Include/
│   │   └── Libraries/
│   └── Rustmt5-chart/
├── feature_process.png
├── LICENSE
├── metrics.png
├── requirements.txt
├── STACKS.md               
└── sractch.md

Usage

For detailed technical documentation, please refer to STACKS.md.

Quick Start (Python)

1. Install dependencies:

   pip install -r requirements.txt
   

2. Run the full pipeline:

   python python_version/main.py run
   

Quick Start (Rust)

1. Build the project:

   cd rust_ml_trader
   cargo build --release
   

2. Execute backtest:

   cargo run --release
   

Citation

If you use this repository in your research or project, please cite it as follows:

@misc{albeos2026ml3mtrader,
  author = {Rembrant Oyangoren Albeos},
  title = {ML-3m-trader: XAUUSDc 3-Minute Timeframe ML Trading System},
  year = {2026},
  publisher = {Hugging Face},
  journal = {Hugging Face Repository},
  howpublished = {\url{https://huggingface.co/algorembrant/ML-3m-trader}}
}