yichengyang-ethan/oracle3/pm-quant-strategy-authoring

PM Quant Strategy Authoring

Use this skill when the user asks the agent to come up with a strategy and write the code.

Goals

• A runnable strategy class (a QuantStrategy subclass).
• Exposed numeric parameters, so the strategy can be auto-tuned.
• strategy validate and backtest run both succeed.

Code entry points

• Base class contract: oracle3/strategy/quant_strategy.py (QuantStrategy).
• For LLM-driven strategies use AgentStrategy instead (see pm-agent-strategy-authoring).
• Example strategies: examples/strategies/*.py, strategies/*.py.
• New strategies go in strategies/.

Steps

1. Create the skeleton

• oracle3 strategy create --output strategies/<name>.py --class-name <ClassName> --type quant

1. Implement the strategy

• In process_event, handle only the event types you need (usually PriceChangeEvent).
• Put parameters in the constructor; numeric parameters are what tuning searches over.
• Call trader.place_order(...) to act.
• Record actions and signals with self.record_decision(...).

1. Quick validation

• oracle3 strategy validate --strategy-ref strategies/<name>.py:<ClassName> --strategy-kwargs-json '<json>' --dry-run --events 10 --json

1. Single-market backtest

• oracle3 backtest run --history-file <history.jsonl> --market-id <M> --event-id <E> --strategy-ref strategies/<name>.py:<ClassName> --strategy-kwargs-json '<json>' --json

1. Prepare for tuning

• Every parameter has a clear meaning, bounds and a default.
• Parameter types stay JSON-serializable.
• Tune with oracle3 research auto-tune --param-grid-json '<json>' ....

Hard rules

• Do not hard-code strategy logic in a command-line script; it belongs in its own strategy file.
• Never use future information (no look-ahead).
• Keep runs reproducible: record the strategy file path, class name, kwargs and command.