Neuro Future
- Indicators
-
Sergey Rozhnov
About me: I have experience in Forex trading and software development since 2006.
I also accept individual orders for the development and improvement of Expert Advisors or indicators in MQL5 and MQL4 languages on a Freelance.
Please contact me via private messages for any inquiries. - Version: 2.6
- Updated: 25 May 2026
- Activations: 5
Neural Network Indicator with Built-In Learning.
This is a real autonomous artificial intelligence system powered by a multi-level customizable perceptron and advanced training settings to predict future price changes. You can train the indicator yourself on any finansial symbol, everything you need is collected in this tool and conveniently implemented with automatic mode and presets.
Major update v2.0 - 2.5 has been released. Advanced input features: Added support for indicators. Now you can combine raw price data with three technical indicators for a deeper market analysis, or use only indicators without prices.
Non-Blocking Live Training: The intelligent training algorithm runs asynchronously in the background. Your MetaTrader 5 terminal remains fully responsive, allowing you to continue trading and analyzing while the neural network is learning.
- Train Directly in MT5. You have a unique opportunity to train the network within the terminal on any symbol and timeframe without using other programming languages. The latest MT5 compiler handles the learning process efficiently.
- Advanced Forecasting. Get a forecast for a selected number of future bars with a customizable notification system.
- Smart File Management. A convenient system for saving neural networks to files with auto-naming and auto-loading for selected charts and periods. This eliminates the need to enter neural network names manually when loading the indicator.
- Validation & Early Stopping. The indicator includes a validation system (with Profit, Binar, and NetError criteria) and an early stopping function. It reserves a portion of the latest history to test its neural network, checking predictions on future bars. This allows you to determine the success of training and protect the network from overfitting. The indicator will stop automatically when further training is ineffective. (To activate validation, set Validation period >0 (for example 200), Use validation criteria for early stopping = true, Save best weights = true).
- Easy Start. Simple settings for a quick start with Auto mode. (For example: Set Max training epochs=5000, Max training samples=1500 and the validation settings as above, then click the learning button).
- Advanced Customization. For advanced users, several types of neural networks and activations are available, along with preset combinations for easy operation. You can configure neural network layers, activations, gradients, and more. Manually stop and continue with buttons learning as needed.
- EA Integration. Easily integrate the indicator into your Expert Advisor.
