Skip to content

README.md

Latest commit

 

History

History
38 lines (24 loc) · 1.66 KB

README.md

File metadata and controls

38 lines (24 loc) · 1.66 KB

MetaTrader 5 Machine Learning Blueprint – Modules Repository

This repository contains the companion code modules, utilities, and assets for the MetaTrader 5 Machine Learning Blueprint article series by Patrick Murimi Njoroge.
It is designed to be a clean, production‑ready implementation of advanced financial machine learning techniques — from robust data handling to adaptive, probabilistic trade execution.

📚 Series Context

This repo accompanies the following articles:

  1. Part 1 – Data Integrity & Tick‑Based Bars

    • Eliminating data leakage with proper tick aggregation
    • Timestamp correction and unbiased dataset preparation

  2. Part 2 – Meta‑Labeling & Triple‑Barrier Method

    • Risk‑aware labeling with profit‑taking/stop‑loss logic
    • Meta‑labels to improve classifier precision under realistic trading constraints

  3. Part 3 – Advanced Labeling & Sample Weighting

    • Trend‑scanning labels with adaptive horizons
    • Purged cross‑validation
    • Sample weighting to address concurrency bias
    • Probabilistic position sizing for real‑time execution

🔑 Key Features

  • Leakage‑Proof Labeling – Triple‑barrier & adaptive trend‑scanning with volatility regime filtering

  • Numba‑Accelerated – 100×–350× faster execution for live‑trading scenarios

  • Concurrency‑Aware Weighting – Down‑weights overlapping observations for better generalization

  • Probabilistic Position Sizing – Trade sizing aligned with model confidence and risk parameters

  • MT5 Integration – Direct pipeline from Python model output to MetaTrader 5 execution