OVERRIDES runtime profile: uses direct_python (plain Python, no model) because projects/crypto-trader/scripts/score_verdicts.py only scores already-recorded Opus verdicts against price outcomes (Postgres + price API reads) — the verdicts themselves are produced by the separate crypto-trader-cycle task; no claude CLI, no LLM API (MC-4942 U12 sweep).
Daily scoring pass for crypto trader Opus verdicts.
Runs at 07:40 SAST (05:40 UTC), after the morning Opus call at ~04:00 UTC and after the swing trader scorer at 05:15 UTC.
Fetches every verdict from opus_verdicts (Postgres crypto_trader DB) and scores directional accuracy: did BTC move in the direction the verdict suggested (long/short) over the next 24 hours and 7 days?
Writes to mc.db:crypto_verdict_scores. Mission Control /trading reads
this and shows per-bias hit rate, avg return, and a recent verdicts feed.
Deduped by day in the UI (one row per date → latest verdict per day). Rebuild from scratch: rerun with --backfill (safe, idempotent).
cd /home/lucienne/workspace && python3 projects/crypto-trader/scripts/score_verdicts.py
Format: minute hour day month weekday
| Started | Status | Duration | Log |
|---|---|---|---|
| 2d ago | completed | 2s | View log |
| 3d ago | completed | 2s | View log |
| 4d ago | completed | 2s | View log |
| 5d ago | completed | 2s | View log |
| 6d ago | completed | 2s | View log |
| 7d ago | completed | 2s | View log |
Mission Control is the board for your delegated work: requests come in, Luci coordinates the next step, and evidence stays visible for review.
Luci is your always-on assistant for routing, status updates, and follow-through. Operators can still open deeper evidence when needed.