FAIRLINE — Tonight Plan (Test Tick)

ROI: QUALITY

This is a proof-of-pipeline tick for Neon (FAIRLINE).

3 highest-leverage code tasks for tonight

1) Tighten the Edge Board into a true ‘terminal’ (fast, scannable, auditable)

Why it matters: makes FAIRLINE feel like premium intelligence software and reduces decision fatigue. Likely files:

• src/app/events/page.tsx
• src/app/events/fairline.css
• src/app/events/PredictionsDrawer.tsx / PredictionsModal.tsx DoD:
• Board shows 5–7 max candidates (or fewer) with clear “why surfaced” tags.
• Every card has: market, books count, movement window, EV/edge, confidence/quality.
• Research mode remains accessible without board gating.

2) Movement intelligence sanity + persistence checks (Neon Postgres)

Why it matters: movement windows must reflect real historical snapshots, or the whole ‘intelligence’ story collapses. Likely files:

• src/lib/pgOdds.ts
• src/lib/scan.ts
• src/app/api/scan/run/route.ts DoD:
• For any candidate, movement 30/60/180m is derived from persisted snapshots.
• If insufficient history, show explicit “insufficient movement history” flag (no guessing).

3) Audit-first research ingest improvements (snippets + links only)

Why it matters: builds trust + keeps the system legally/ethically safe. Likely files:

• src/app/api/context/ingest/route.ts
• src/lib/context.ts
• src/app/api/audit/event/route.ts DoD:
• Each ingest run stores: source URL, snippet, published_ts_ms (null if unknown), mapping confidence.
• If ESPN mapping confidence below threshold → do not write ESPN IDs.

Safety reminders (non-negotiable)

• No pick lists / no “top plays.”
• Decision-support + audit artifacts only.
• Research ingest is snippets + links only (no full article copying).

Next action

• Run a real FAIRLINE tick (non-test) with a single selected objective (pick one of the three above).