FreqAI-Strategy is an advanced AI-driven trading strategy built for Freqtrade using Long Short-Term Memory (LSTM) neural networks. This strategy is designed to:

• Predict future price trends in cryptocurrency markets using deep learning.
• Execute trades based on AI-generated signals to maximize profitability.
• Adapt dynamically to market conditions using engineered features.

🚨 Work in Progress: This strategy is still under active development and is not meant for live trading with real money. Use it for research and backtesting only.

✅ Uses LSTM for time-series forecasting
✅ Dynamic target scaling and market regime filtering
✅ Backtesting and hyperparameter tuning support
✅ Supports multiple timeframes (1h, 2h, 4h)
✅ Automated model training and retraining ✅ Optimized for training on GPU/CPU ✅ Optimized for Binance Futures Trading

# Clone and install Freqtrade
git clone https://github.com/freqtrade/freqtrade.git
cd freqtrade
./setup.sh --install

pip install -r requirements.txt

git clone https://github.com/GoodyNick/Freqai-Strategy.git
cd Freqai-Strategy

After cloning, ensure the configuration, strategy, and model files are correctly placed in the Freqtrade directory structure. Use the following commands:

# Copy configuration file
cp config-torch-lstm_v2.json /freqtrade/user_data/configs/

# Copy strategy file
cp ExampleLSTMStrategy_v2.py /freqtrade/user_data/strategies/

# Copy model-related files
cp PyTorchLSTMModel_v2.py /freqtrade/freqtrade/freqai/torch/
cp PyTorchLSTMRegressor_v2.py /freqtrade/user_data/freqaimodels/
cp PyTorchModelTrainer_v2.py /freqtrade/freqtrade/freqai/torch/
cp freqai_interface.py /freqtrade/freqtrade/freqai/

Modify them as needed before running Freqtrade.

Modify config-torch-lstm_v2.json to customize:

• Train/Test periods (train_period_days, backtest_period_days)
• Feature Engineering Parameters (DI threshold, scaling methods)
• LSTM Model Parameters (hidden_dim, num_lstm_layers, dropout, etc.)
• Trading Settings (Max trades, margin mode, stake size)
• model training parameters
• ** ... **

freqtrade backtesting --config config-torch-lstm_v2.json --strategy ExampleLSTMStrategy_v2 --freqaimodel PyTorchLSTMRegressor_v2 

you can also use run.sh script for backtesting, plotting, or hyperopt freqai strategy

🤝 Contributions are welcome! If you have suggestions or improvements, feel free to submit a pull request or open an issue.

📬 Contact: GitHub Issues or reach out on Discord!

License: MIT