Run artifacts

Language: English · 简体中文

strategy_pipeline.pipeline.output_artifacts writes configured datasets, features, signals, evaluation results, backtest outputs, and diagnostics under a run directory. write_run_artifacts(context=...) returns a mapping of artifact names to paths or summaries.

from strategy_pipeline.pipeline.output_artifacts import write_run_artifacts

artifacts = write_run_artifacts(context=context)

The module handles file formats, paths, and the artifact-path inventory. Owner packages provide data access, signal calculation, portfolio rules, and execution models. Backtest and position artifacts use public portfolio-backtester interfaces; signal artifacts use the public quant-market-research interface.

The caller supplies the run context, and configuration controls which outputs are written. The function does not read credentials, generate strategy signals, or change portfolio logic.

An owner may provide weekly_basket_performance_path. The pipeline then records that path in the returned artifact inventory; it does not read the file, scan the run directory, calculate performance, or import strategy code. If omitted, the field is None.