feedback-loop

name	feedback_loop
description	Post-trade feedback loop system: shaped-reward logging, diagnostics, and self-heal. Use when investigating trade outcomes, checking system health, or understanding automated corrective actions.
user-invocable	false

Post-Trade Feedback Loop

Automated post-trade analysis pipeline that logs shaped rewards, runs health diagnostics, and applies corrective actions without human intervention.

---

Loop Diagram

Trade Close (OANDA)
       |
       v
PostTradeLoop.run(trade_result)
       |
       +---> [1] Idempotency check (trade_id dedup)
       |
       +---> [2] Load journal entry from trade_journal_rl.json
       |
       +---> [3] Compute shaped reward
       |
       +---> [4] Append to logs/trade_feedback.jsonl
       |
       +---> [5] Append learning if |reward| > 2R
       |
       +---> [6] PostTradeDiagnostics.run()
       |            |
       |            v
       |     Structured health report
       |     {status: HEALTHY|DEGRADED|CRITICAL}
       |            |
       |            v (if not HEALTHY)
       +---> [7] SelfHeal.apply(diag_result)
                     |
                     v
              Decision log + corrective actions

---

When It Fires

• Per trade close: Hooked into close_trade_oanda() in src/scanner/execution.py. Every confirmed OANDA trade close triggers PostTradeLoop.run().
• Per cycle: The orchestrator dispatches feedback processing each scan cycle via run_cycle().
• Manual: Via MCP tools (trigger_rl_update, trigger_diagnostics, ) or CLI ().