docs: Add technical and how-to-run documentation

howtorun.md

@@ -0,0 +1,44 @@
+# How to Run
+
+## Prerequisites
+
+*   MetaTrader 5 terminal installed.
+*   Python 3.x installed.
+*   The following Python libraries installed: `pandas`, `scikit-learn`, `skl2onnx`, `onnxruntime`.
+
+## Installation
+
+1.  Clone the repository to your local machine.
+2.  Install the required Python libraries using pip:
+
+    ```bash
+    pip install pandas scikit-learn skl2onnx onnxruntime
+    ```
+
+## Running the project
+
+1.  **Export Price Data**:
+    *   Open the MetaEditor in your MetaTrader 5 terminal.
+    *   Open the `Export_EURUSD_History.mq5` file.
+    *   Compile the script.
+    *   Run the script on a EUR/USD chart. This will create a `raw_price_data.csv` file in the `MQL5/Files` directory of your MetaTrader 5 installation.
+    *   Copy the `raw_price_data.csv` file to the root of the project directory.
+
+2.  **Train the Models**:
+    *   Run the `create_benchmark_model.py` script to create the benchmark model.
+
+        ```bash
+        python create_benchmark_model.py
+        ```
+
+    *   Run the `Model_Development.py` script to train the neural network model.
+
+        ```bash
+        python Model_Development.py
+        ```
+
+3.  **Run the Expert Advisors**:
+    *   Copy the `Benchmark_Model_EA.mq5` and `Neural_Network_Trader_EA.mq5` files to the `MQL5/Experts` directory of your MetaTrader 5 installation.
+    *   Copy the `benchmark_logistic_model.onnx` and `trading_neural_network_model.onnx` files to the `MQL5/Files` directory of your MetaTrader 5 installation.
+    *   Open the MetaEditor and compile the EAs.
+    *   Attach the EAs to a EUR/USD chart in your MetaTrader 5 terminal.

tech.md

@@ -0,0 +1,34 @@
+# Technical Documentation
+
+## Project Overview
+
+This project is a trading bot for MetaTrader 5 that uses a machine learning model to make trading decisions. The project consists of MQL5 scripts for interacting with the MetaTrader 5 terminal and Python scripts for developing and training the machine learning model.
+
+## Technology Stack
+
+*   **MQL5**: Used to create the Expert Advisors (EAs) that run in the MetaTrader 5 terminal.
+*   **Python**: Used for data analysis, machine learning model development, and training.
+*   **Pandas**: Used for data manipulation and analysis.
+*   **scikit-learn**: Used for creating and training the machine learning models.
+*   **ONNX**: Used to export the trained models to a format that can be used by the MQL5 EAs.
+*   **skl2onnx**: Used to convert scikit-learn models to ONNX format.
+
+## File Descriptions
+
+*   `Export_EURUSD_History.mq5`: An MQL5 script that exports historical price data for the EUR/USD pair to a CSV file.
+*   `Model_Development.py`: A Python script that loads the exported price data, performs feature engineering, and trains a neural network model. The trained model is then exported to an ONNX file.
+*   `create_benchmark_model.py`: A Python script that creates and trains a simple logistic regression model to be used as a benchmark. The model is exported to an ONNX file.
+*   `Benchmark_Model_EA.mq5`: An MQL5 Expert Advisor that uses the benchmark logistic regression model to make trading decisions.
+*   `Neural_Network_Trader_EA.mq5`: An MQL5 Expert Advisor that uses the trained neural network model to make trading decisions.
+*   `raw_price_data.csv`: A CSV file containing historical price data for the EUR/USD pair.
+*   `benchmark_logistic_model.onnx`: The exported benchmark logistic regression model in ONNX format.
+*   `trading_neural_network_model.onnx`: The exported trained neural network model in ONNX format.
+*   `MT5-PY-AI-Tbot`: A file containing the concatenated code of all the project files.
+*   `separate_codes/`: A directory containing the individual project files.
+
+## Model
+
+The project uses two machine learning models:
+
+1.  **Benchmark Model**: A simple logistic regression model that predicts the direction of the next day's price change based on the previous day's price change.
+2.  **Neural Network Model**: A multi-layer perceptron (MLP) classifier that predicts the direction of the next day's price change based on the previous day's price change. The model is trained using hyperparameter tuning with `RandomizedSearchCV` and `TimeSeriesSplit` to find the best parameters.