Scan And Frame

Research-tier idea generation with regime framing. Chains market-regime (context) + universe-builder (candidates) + relative-strength (rank candidates vs SPY). Optionally adds factor-research for broader factor context (heavy, off by default). Different from portfolio-review (positions) and weekly-brief (macro-only) — this is discovery mode.

rgourley 761ee69 5 files · 5.4 KB Updated

File contents

scan-and-frame

Discovery-mode composite. Runs universe-builder with the operator's filters, ranks the top N by relative strength vs SPY, and frames the whole thing with the current market regime. Optional heavy factor- research pass for factor context.

When to invoke

  • Analyst says "what should I look at right now", "find me candidates in X sector", "screen for momentum names"
  • Weekly / periodic universe scan with regime context
  • Different from portfolio-review (positions) and stock-one-pager (retail single-name)

Modes

  • Fast (default): market-regime + universe-builder + relative- strength. Runs on Stocks Starter, ~10-30s depending on universe size.
  • Full (include_factor_research=True): also runs factor-research. Heavy — 3-year factor panel over 200-name universe. Use for weekly cadence, not per-run.

What you get back

Layer 1 JSON matching output-schema.json. Headline: regime, universe count, top-5 RS-ranked candidates, top factor (if factor-research ran).

Layer 2 rendered brief. See references/rendering.md.

rgourley/quant-garage/tree/main/skills/scan-and-frame commit 761ee69850

Frequently asked questions

npx skillmds@latest add rgourley/scan-and-frame