AI Trading: Lesson Learned #110: Trailing Stops Script Existed But Never Executed

Source: Dev.to

Problem

A paper account held $1,113 in unrealized profits that were unprotected because the trailing‑stop script was never run.

• scripts/set_trailing_stops.py was created on Jan 7 2026 but never executed.
• The daily-trading.yml workflow contained no step to set trailing stops.

Unprotected Positions (excerpt from data/system_state.json)

{
  "symbol": "INTC260109P00035000",
  "unrealized_pl": 151.0
},
{
  "symbol": "SOFI260123P00024000",
  "unrealized_pl": 56.0
},
{
  "symbol": "AMD260116P00200000",
  "unrealized_pl": 457.0
},
{
  "symbol": "SPY260123P00660000",
  "unrealized_pl": 449.0
}

Total at risk: $1,113 in profits.

Why the Script Was Not Used

• The script existed as a standalone file: scripts/set_trailing_stops.py.
• It was referenced only in claude-agent-utility.yml (manual trigger).
• It was absent from the daily-trading.yml workflow, meaning:
  • Positions were never automatically protected.
  • Manual execution was required each run.
  • A single market gap could erase all profits.

Fix Implemented

Added an automatic step to daily-trading.yml:

- name: Set Trailing Stop-Loss Orders (Phil Town Rule 1)
  if: success()
  env:
    ALPACA_API_KEY: ${{ secrets.ALPACA_PAPER_KEY }}
    ALPACA_SECRET_KEY: ${{ secrets.ALPACA_PAPER_SECRET }}
  run: |
    python3 scripts/set_trailing_stops.py

Core Rules Reinforced

1. Rule #1: Don’t lose money.
2. Rule #2: Don’t forget Rule #1.

Our system was violating both by:

• Holding profits without protection.
• Relying on hope instead of stop‑losses.

Lessons Learned

• A risk‑protection script is useless if it requires manual execution.
• Every risk‑management tool must be integrated into the automated CI/CD workflow immediately after creation.
• Add a verification step to confirm the trailing_stops section exists in system_state.json.

Tags: trailing_stops, risk_management, phil_town, rule_1, automation_gap, ci_integration

This lesson was auto‑published from our AI Trading repository.

More lessons: rag_knowledge/lessons_learned