Skip to content

Latest commit

 

History

History
52 lines (36 loc) · 2.81 KB

README.md

File metadata and controls

52 lines (36 loc) · 2.81 KB
Raw

High-Frequency-Trading-by-Machine-Learning-and-Deep-Learning

This project aims to predict future strategies by applying both machine learning and deep learning techniques. It leverages a variety of models including LSTM (Long Short-Term Memory), ALSTM (Attention-based LSTM), TCN (Temporal Convolutional Networks), GATS (Graph Attention Networks), and SFM (State Frequency Memory) to analyze and predict based on time-series data.

Structure

The project is organized into several Python scripts, each serving a specific purpose:

  • utils.py: Contains utility functions for preprocessing and manipulating data.
  • LSTM.py: Defines the LSTM model for time-series prediction.
  • ALSTM.py: Implements the ALSTM model, which enhances LSTM with an attention mechanism.
  • TCN.py: Contains the Temporal Convolutional Network model, suitable for sequence modeling tasks.
  • GATS.py: Defines the Graph Attention Network model, leveraging attention mechanisms for graph-structured data.
  • SFM.py: Implements the State Frequency Memory model for capturing multi-frequency patterns in time-series.
  • ML.py: Provides implementations of various machine learning models for comparison and ensemble learning.
  • DL.py: Central script that orchestrates the data loading, model training, evaluation, and prediction tasks using deep learning models.

Features

  • Preprocessing: Standardizes and transforms the input data to be model-ready.
  • Model Implementation: Offers a diverse set of models to tackle time-series prediction from different angles.
  • Evaluation and Prediction: Supports model training, performance evaluation, and future strategy predictions.
  • Extensibility: Easily extendable framework to include more models or adapt existing ones to new datasets.

Usage

To use this framework for your prediction tasks, follow these steps:

  1. Prepare your dataset according to the format expected by the utils.py preprocessing functions.
  2. Choose the model(s) you wish to train from the list of available Python scripts (e.g., LSTM.py, ALSTM.py).
  3. Use the DL.py script to train and evaluate the model on your dataset. This script will automatically handle data loading, model initialization, training, and evaluation.
  4. For machine learning models, use the ML.py script to perform similar tasks with classical machine learning approaches.

Dependencies

This project requires the following Python libraries:

  • NumPy
  • Pandas
  • PyTorch
  • scikit-learn
  • Matplotlib (for data visualization)
  • tqdm (for progress bars)

Ensure these are installed in your Python environment before running the scripts.

Contribution

Feel free to contribute to this project by suggesting improvements, adding new models, or optimizing existing code.

License

This project is licensed under the MIT License. See the LICENSE file for details.