`.yoloforever/ledger.md`

9 lines
## Cycle 2 — 2026-08-28
- **Pick (Cody-seeded):** validate the buy-score on the small/mid-caps the tool actually scans (backtest was large-cap only).
- **Finding:** the score is CAP-DEPENDENT — large-cap 120d corr +0.68 (works), small-cap −0.6 (INVERTS). Decision-changing.
- **CLARIFY (Steve present):** chose "segment backtest by cap + warn."
- **Shipped:** segmented backtest (large vs small columns), per-stock cap-tier flag (large=sweet-spot, <$20B=INVERTS, $20–40B=untested), /api/cap for manual adds.
- **Gate:** panel spec 15/15, 0 console errors, tier flags verified (PLUG→INVERTS, SOFI→untested), prod green.
- **Next seed:** propagate cap-tier awareness to SCAN RESULTS — flag small/mid-cap matches with ⚠ + add a "large-caps only" scan option, so users screen the validated universe.