Auto-generated. This article is rebuilt from FastAPI route introspection by scripts/build_api_kb.py. Edit the route docstring in app/ and re-run the script — do not edit this file directly.

GET /api/v1/screener/board-track-record

Handler: app.screener.routes.get_board_track_record
Methods: GET

Description

Board-pick track record — grades the board’s ACTUALLY-SURFACED picks.

Reads screener_board_picks (populated by each screener run snapshot), fetches universe price bars and SPY closes, then aggregates:

• closed — matured picks (horizon elapsed): overall directional hit-rate
  • mean/median signed excess return, broken out by side / horizon / archetype / regime-band. hit_rate is a 0-1 fraction. Carries the W1-8/F16 raw-only split: n_market_adjusted vs n_raw_only — a pick graded with no SPY bar for its window matured on its RAW return (beta-contaminated), and the split makes that mix visible instead of silently pooling it.
• open — picks still inside their horizon: count + mean MTM signed excess.
• counts — number of run snapshots captured + total picks recorded.
• caveats — honesty surface: warming_up (closed.n < readability floor), forward_only_note, honesty_note.
• contracts — PR-7 (Scheme 4) the STRICTER lens: the LOCKED pick contracts (screener_pick_contracts) graded held-to-their-ACTUAL-exit, market-adjusted vs SPY. Self-versioned (contracts.schema_version); additive to the board-picks lens above. Carries credibility (warming < 10 matured / early 10-29 / on_record >= 30 — a DISPLAY label that licenses no sizing), open (per-position paper MTM-vs-SPY + mean_paper_excess), closed (the held-to-exit record with by_exit_reason — horizon_elapsed / invalidated_price / invalidated_regime — plus by_side and the raw-only disclosure), and counts (open / closed / exited_total). Where the board-picks lens grades every DISPLAYED pick at a fixed ladder, this grades only the positions the board actually COMMITTED to and held.

Cold-start (forward-only)

Board weights + regime tilt are calibration-state at snapshot time and cannot be reconstructed historically. Picks accrue FORWARD from deploy; the first graded results appear after 5-10 trading days. While warming_up is True the byline must render “building” — never display a fabricated track record.

PUBLIC — market data only (no PII, no per-account data). Degrade-honest per DOCTRINE P0: any internal error returns HTTP 200 with an empty-but-valid payload, never HTTP 500.

Curl

curl -s https://bigclawd.com/api/v1/screener/board-track-record