Strategy Deployer

Deploy trading strategies with risk gating, monitoring, and guarded live promotion.

SKILL: strategy-deployer

Purpose

Deploy trading strategies safely using an explicit lifecycle: validate → risk-check → paper deploy → monitor → (guarded) live promote → scale/stop.

When to Use

• A strategy is ready after backtesting.
• Moving from research to paper trading.
• Promoting a paper-tested strategy to live trading (guarded).

Inputs

• strategy_spec (required, object|string): rules, signals, markets, timeframes.
• mode (optional, enum: paper|live, default: paper)
• risk_limits (required, object): max risk per trade, max drawdown, exposure caps.
• validation_artifacts (optional, object): backtest report, paper performance stats.
• approval (optional, string): explicit user/Morpheus approval reference for live.

Steps

1. Verify prerequisites:
   • strategy has a written hypothesis and defined invalidation conditions
   • backtest/paper artifacts exist (as applicable)
2. Run risk checks:
   • enforce risk per trade
   • enforce exposure caps
   • enforce max drawdown thresholds
   • ensure kill-switch is configured
3. Deploy to paper (default):
   • start with small allocation
   • emit logs for every signal and order
4. Monitor and record:
   • trades
   • slippage
   • drawdown
   • regime notes
5. Promote to live only if:
   • explicit mode=live
   • explicit approval is present
   • preflight checks pass
6. Define rollback/stop rules and execute them on trigger.

Validation

• Paper mode produces complete logs and trade history entries.
• Live mode is blocked unless approval + flags are present.
• Risk constraints are enforced before every order.
• Stop conditions are tested (kill-switch works).

Output

deployment: "paper|live"
strategy: "<name/id>"
markets: ["..."]
risk_limits: { "...": "..." }
status: "deployed|blocked|stopped"
audit_events: ["signal_detected", "risk_check_passed", "order_submitted", "order_filled", "trade_recorded"]

Safety Rules

• Never bypass risk checks to “capture opportunity”.
• Never deploy live without explicit approval and a paper validation phase.
• Never imply profit certainty; report performance as historical/conditional.

Example

Paper deploy:

• strategy_spec: “Mean reversion in range regime, 5m timeframe.”
• mode: paper
• Output: deployment report + monitoring checklist + stop triggers.