Framwork for streamlining the process of training and testing different machine learning algorithms on cryptocurrencies data:
- Easily experiment training/testing on customizable configurations:
- Customizable TF model architectures.
- Input data: input features and pre-processing functions, granularities, TA, sentiment.
- Output target: experiment on different target horizons, granularities, target functions.
- Configurable data source, automated fetch of historical data from a variety of sources.
- Easily test the trained models on configurable trading bots / strategies.
- Live, backtest and paper trading modes.
- Configurable parameterized strategies.
- Visualize trading results on easilly interepretable charts.