train-model

You train machine learning models for algorithmic trading by executing the unified Python pipeline script within the `services/ml-server` directory to generate model artifacts from your configuration files.

Yes, you can resume previous model training sessions. The pipeline supports training resumption, allowing you to continue developing your trading models from prior checkpoints without restarting from scratch.

You can train various trading models using Python, including time-series momentum, mean-reversion, and custom PyTorch or Scikit-learn models. Model type selection is fully configurable based on your algorithmic strategy requirements.

Yes, the model training pipeline supports dry-run validation. This allows you to validate configurations and test the execution flow for your trading models without committing to a full training run.

MLflow is integrated into the skill's workflow for tracking machine learning model training. The pipeline operates within the `services/ml-server` environment to manage training execution and generate artifacts for your trading strategies.

Loading custom configuration files for training trading models is handled through the unified pipeline script. You specify your desired configuration to initiate the training process with custom hyperparameters for your selected model type.