# Expert/ Directory Documentation

## ExpertCustom.mqh
- **Purpose:** Defines the `CExpertCustom` class, a base class for custom expert advisors in the EA framework.
- **Inheritance:** Inherits from `CExpert` (not shown here, likely a core MQL5/MetaQuotes class or defined elsewhere in the project).
- **Key Features:**
  - Provides virtual event handlers for MQL5 events: `OnTick`, `OnTimer`, `OnChartEvent`, etc., allowing derived classes to implement custom trading logic.
  - Contains methods for initializing trading objects, handling series data, and managing trading events.
  - Designed for extensibility: users can derive their own expert logic by subclassing and overriding the provided virtual methods.
- **Modernization Note:**
  - This class is a good candidate for further abstraction to support AI/ML-driven strategies. Consider introducing interfaces or abstract base classes for signal generation, money management, and risk control, allowing plug-and-play of traditional and AI/ML modules.
  - Ensure all event handlers are unit-testable and decoupled from hard-coded logic.

## ExpertMoneyCustom.mqh
- **Purpose:** Implements a custom money management class (`CExpertMoneyCustom`) for the EA, extending `CExpertMoney`.
- **Key Features:**
  - `CheckAndCorrectVolumeValue`: Ensures trade volume is within broker constraints (min/max/step), adjusting and describing corrections.
  - `CheckAndAdjustMoneyForTrade`: Dynamically adjusts lot size to fit available margin, reducing lots if margin is insufficient, and handles errors gracefully.
- **Modernization Note:**
  - This class is a good candidate for AI/ML-driven position sizing. Consider abstracting the logic to allow for ML-based risk and lot size optimization.
  - Ensure all adjustments are logged for transparency and auditability.

## ExpertSignalAIBase.mqh
- **Purpose:** Provides a base class (`CExpertSignalAIBase`) for AI/ML-driven signal generation, extending `CExpertSignalCustom`.
- **Key Features:**
  - Integrates with the neural network/AI subsystem (`Network.mqh`).
  - Manages a wide range of indicator objects (Open, Close, High, Low, Volumes, AD, ADX, MACD, etc.) for feature extraction.
  - Supports dynamic configuration of neural network topology, training, and feature selection.
  - Implements methods for indicator initialization, data buffering, training, and topology persistence.
- **Modernization Note:**
  - This is the main integration point for advanced ML/AI logic. Ensure all feature selection and training parameters are externally configurable.
  - Refactor to support plug-and-play feature pipelines and automated hyperparameter optimization.

## ExpertSignalCustom.mqh
- **Purpose:** Implements a custom signal logic class (`CExpertSignalCustom`) for the EA, extending `CExpertSignal`.
- **Key Features:**
  - Adds database-driven signal tracking, pattern recognition, and trade record management.
  - Supports dynamic filter addition, ATR-based entry/exit logic, and advanced signal buffering.
  - Implements robust duplicate detection, trade status management, and event-driven signal processing.
- **Modernization Note:**
  - This class is central to integrating traditional and AI/ML signals. Refactor to decouple hard-coded logic and support dynamic, testable signal pipelines.
  - Ensure all database operations are abstracted for testability and future migration to more advanced data stores.

---

(Other files in Expert/ will be documented as they are processed.)