AI Algorithmic Trading Solution

FxMath Lumina Major Update: Version 1.12 Launches with Genetic Optimizer, Paper Trading, and Multi-Instance MT5 Support

Algorithmic trading demands continuous adaptation as market regimes shift. We are excited to announce a major upgrade for FxMath Lumina, moving the platform from version 1.02 to v1.12. This release transforms FxMath Lumina into a highly versatile, GUI-based MT5 expert advisor capable of trading any instrument and timeframe with built-in optimization, backtesting, and multi-instance execution.

“FxMath Lumina is now capable of trading any pair and timeframe on the MT5 platform, and you can run multiple instances simultaneously. This GUI-based version is the most advanced EA I have developed to date — it also includes built-in backtesting and optimization features.”
— Reza Mehrabani, Lead Developer

What is FxMath Lumina?

FxMath Lumina is a standalone Windows application and MT5 integration powered by a robust Python + LightGBM machine learning engine. It allows traders to generate and execute institutional-grade machine learning strategies on any market (Forex, Commodities, Indices, Crypto) without writing a single line of code.

  • LightGBM AI Engine: Lumina learns the optimal entry and exit rules based on your target Risk-to-Reward (R:R), chosen asset, and preferred timeframe.
  • Walk-Forward Validator: Out-of-Sample (OOS) testing verifies the strategy’s integrity before deploying real capital.
  • Parallel Trading: Run 10 to 50 unique strategies simultaneously as separate instances on your MT5 terminal.
  • Historical Performance Baseline: OOS Win Rate: ~68% | Profit Factor: 2.84 | Max Drawdown: 12.3% | Sharpe Ratio: 1.87.

Key Architectural Upgrades in the v1.12 Release

1. Built-in Genetic Optimizer (GA) Tab

Traders can now optimize up to 11 parameters directly from the user interface. Powered by both grid search and genetic algorithms, the engine automatically detects your active symbol, sweeps through parameter grids (including the new price_ahead variable), filters out sparse configurations, and saves the highest-performing configurations straight to your settings panel.

2. Fully Simulated Paper Trading & Logging

Test your strategies live under actual market conditions without risking a single dollar. The newly integrated Paper Trading mode operates entirely with in-memory simulated positions, writing transaction details to a CSV-based local journal backed by secure SHA-256 configuration hashes to verify strategy consistency.

3. Multi-Instance MT5 Scalability

We refactored the monolithic 2,697-line engine into a modular, production-ready fxmath_lumina/ package. This allows for stable, low-latency execution of unlimited simultaneous instances. Each instance is assigned its own unique Magic Number and runs autonomously with individual trailing stop and breakeven configurations.

4. Interactive HTML Performance Reports

Lumina now generates comprehensive HTML performance reports that open automatically in your web browser. These reports include a visual equity curve, hourly P&L distributions, a metrics grid, and detailed transaction history logs.

Complete Development History: v1.02 → v1.12

The progression of FxMath Lumina represents systematic upgrades in risk management, usability, and execution speed:

v1.12 — Paper Trading, Journaling, & Configuration Refactoring
• Extracted runner.py from ui.py (handles charts, HTML reports, and CSV loading)
• journal.py: Added trade CSV logging with SHA-256 configuration hashing
• Paper Trading Mode: Fully simulated, in-memory positions
• Live instance rows now show indicator variables
• Optimizer defaults to Genetic Algorithm (GA) with Min Trades filter to reject sparse configurations
• Auto-detects symbols on optimizer tab load
• Auto-saves best configurations upon completion or manual stop
• Added save/load dropdown for named configurations via best_configs/ folder

v1.11 — Optimizer Polish
• Added Label Method dropdown to the optimizer
• Auto-symbol detection implemented in the optimizer tab
• Added price_ahead parameter to the parameter grid
• Trend filter defaults to ON across all code paths
• Added Clear Log buttons on both the Run Log and Optimizer tabs
• Cast fixes for float/int optimizer values
• Breakeven and Trailing activation percentages use safety-casted integer parameters

v1.10b — Optimizer Tab & UI Polish
• Optimizer Engine: Parameter class, compute_fitness(), grid search, and GA
• Optimizer GUI: Controls 11 parameters, population size, and generations
• Added Start/Stop controls with a progress bar and ranked results table
• “Load Best” button writes parameters directly to the settings panel
• Live Trading UI: Scaled up font sizes, widened log, and cleaned fields for better visibility
• Added Magic Number field in the Add/Edit Instance dialog

v1.10 — Package Refactoring & HTML Reporting
• Monolithic rf_trading.py (2697 lines) split into modular fxmath_lumina/ package
• Auto symbol detection via MT5 connection
• Configurable Trend Filter (SMA/EMA, period, on/off toggles)
• Added Label Method selector (dual_scenario / price_direction) + price_ahead variable
• Interactive HTML Export Report: Equity curve, hourly P&L, metrics grid, trade table
• Automatic browser report launching and updated time formatting
• Integrated 25 unit tests covering indicators, features, and backtesting paths

v1.02 — Risk Management & Machine Learning Regularization
• LightGBM regularization parameters (L1/L2, subsample, feature sampling, tree limits)
• Early stopping implemented with 75/25 validation split
• 3-seed ensemble for stable prediction weights
• Time-decay sample weighting to prioritize recent data
• 15+ technical features (volatility regime, BB Squeeze, Donchian Channels, session, etc.)
• P&L contract calculation fix: CONTRACT_SIZE=100 (1 lot = 100 oz)
• Optional, configurable Breakeven stop and Trailing stop
• Integrated GUI toggles for BE/Trailing in settings and AddInstance dialogs

How the Lumina Workflow Operates

  1. Define Strategy Parameters: Set your target instrument, timeframe, and ideal R:R parameters within the clean GUI dashboard.
  2. Validate Locally: Run the Walk-Forward validator on your broker’s historical data to ensure edge validity before deployment.
  3. Train LightGBM Models: The system trains its machine learning model on your local machine, automatically adapting to structural shifts in volatility.
  4. Deploy & Monitor: Run your models live or in Paper Trading mode. The system manages entries, exits, and trailing protections with precise, one-trade-at-a-time risk parameters.

Secure Your Lifetime License

Get complete, unrestricted access to the standalone Windows GUI application, modular Python engine, and unlimited simultaneous MT5 execution paths.

$299 (Lifetime License — No Recurring Fees)


Get FxMath Lumina Lifetime Access


Risk Warning: Trading financial instruments, including foreign exchange, commodities, indices, and cryptocurrencies, carries a high level of risk and may not be suitable for all investors. Algorithmic performance metrics, backtests, and historical parameters are provided for informational purposes only and do not guarantee future performance. Ensure strict risk management policies are maintained at all times.


Discover more from FxMath Financial Solution

Subscribe to get the latest posts sent to your email.

Discover more from FxMath Financial Solution

Subscribe now to keep reading and get access to the full archive.

Continue reading