This repository contains experiments and implementations related to the Metalabeling strategy applied to both stock and cryptocurrency trading systems.
The figure represents the Metalabeling process, sourced from the master's thesis available at this link.
The project uses two main data folders:
- Data_AAPL: Contains intermediate data for Apple stock trading experiments.
- Data_Crypto: Contains intermediate data for cryptocurrency trading experiments.
The repository includes four Jupyter notebooks, each focusing on different aspects of the trading strategy:
- Apple Trading System Breakout Donchian Channel.ipynb: Implements a trading strategy for Apple stock based on breakout signals using Donchian Channels (primary strategy / model).
- Apple Trading System Metalabeling Gradient Boosting.ipynb: Applies Metalabeling with Gradient Boosting to optimize the trading signals for Apple stock.
- CRYPTO Trading System Donchian Channel.ipynb: Implements a trading strategy for cryptocurrency using Donchian Channels (primary strategy / model).
- CRYPTO Trading System Metalabeling Gradient Boosting.ipynb: Applies Metalabeling with Gradient Boosting to optimize the trading signals for cryptocurrency.
The code for the primary models in these notebooks is primarily derived from the book "Strategie di Trading con Python".
Here are some useful references related to the Metalabeling strategy and trading systems:
- What is Corrective AI?: LinkedIn Article
- What is the Probability of Profit of Your Next Trade? (Introducing predictnow.ai): LinkedIn Article
- Conditional Parameter Optimization: Adapting Parameters to Changing Market Regimes via Machine Learning: LinkedIn Article
- Quantitative Trading: How to Build Your Own Algorithmic Trading Business: Amazon
- Strategie di Trading con Python: Hoepli
- Exploration of a Trading Strategy System Based on Metalabeling and Hybrid Modeling Using the SigTech Platform: Thesis