edge-candidate-agent

To convert EOD OHLCV data into quant trading strategy candidates, you run the auto-detection script on your OHLCV parquet file. It computes features, infers candidate families, and exports validated Phase I-compatible artifacts for pipeline backtesting.

The edge-finder-candidate/v1 interface enforces deterministic schema validation and export rules for Phase I edge research tickets. It ensures your strategy.yaml and metadata.json artifacts remain reproducible and compatible before running expensive pipeline backtests.

Yes, you need Python 3.9+ with YAML support to generate Phase I candidate specs. The PyYAML dependency is required to parse configurations and export validated strategy.yaml files plus provenance metadata.json for your pipeline.

Yes, you can optionally refine edge research tickets using external LLM ideation commands. The workflow allows you to turn anomalies into structured tickets, apply hints or external ideation, and then export validated strategy.yaml artifacts for the pipeline.

The best way to preflight-check interface compatibility is to validate edge-finder-candidate/v1 constraints before running expensive pipeline backtests. This ensures your strategy.yaml and metadata.json files meet deterministic schema export rules for the trade-strategy-pipeline.

Edge candidate research pipelines fail during pipeline handoff when hypotheses cannot be converted into interface-compatible, reproducible tickets and validated artifacts. Enforcing deterministic schema validation and exporting proper strategy.yaml files prevents these downstream backtesting failures.