Talon Weekly — TARGET output format (Glitch/Skylit reference)
The final product PANDORA must be able to generate: Talon's weekly sector watchlist (Glitch, Skylit founder). This doc captures the format/mechanics (not the tickers of a given week) — this is what the V1 scanner must converge toward. Source: Talon Weekly, 28/06/2026.
The Talon pipeline (= our §0.2, validated from the outside)
- Universe scan: ~875 symbols → strict setups (~98) + supplemental VEX/OTE watches (~194) = ~292 output, narrowed into a clean watchlist.
- Setup pulse: bullish vs bearish count, bullish score-weight %, avg snapshot move, sector rotation signal (detected / none).
- Theme clusters (15+): retail, defensive, clean energy, oil majors, AI software, EV, fintech, AI infra, cloud, consumer disc, Mag7, energy services, materials, China ADR, crypto.
- Hedge Dashboard = the THROTTLE: SPXW/SPY/QQQ/SMH/SQQQ/VIX/VXX/TQQQ. The single-name map can be very bullish, but the indices decide the aggressiveness. Confluence: "one warning is not enough; if SPY+QQQ+SMH+SQQQ+VIX+VXX confirm risk-off together → de-risk high-beta longs."
- Per-name setups (primary + secondary).
- Sector rotation read (by theme + leaders).
- Timing framework: immediate · first reaction (0-5d) · swing (2-4 wk) · extended VEX (multi-week/month).
- How to read + What invalidates (nodes evolve, they are not fixed laws).
Setup schema (Talon fields) → maps onto our Setup (§3)
| Talon field | Our field |
|---|---|
| Current | spot |
| Best Entry / OTE | entry (discount/OTE zone) |
| Breakout Hold / Retest | reclaim level |
| Tactical Risk Floor/Ceiling | stop (1 node beyond) |
| Structural Invalidation | invalidation (acceptance beyond) |
| Target Ladder (VEX) | targets[] + target_swing |
| Timeline (first/swing/extended) | setup_type horizon (intraday→swing→VEX) |
| Setup type (strict / VEX·OTE watch / bearish hedge) | verdict GO/WATCH + category |
Hedge Dashboard = our v_joint + touchstone (ALREADY there)
The "the single name is bullish but the indices impose the throttle" IS our v_joint regime (RISK ON/OFF) + the touchstone (GEX/vol = vulnerability → sizes, not direction). Talon formalizes it in 8 control symbols (SPY/QQQ/SMH/SQQQ/VIX/VXX/TQQQ/SPXW) with levels; we have it as the macro read. To enrich: add SMH/SQQQ/VXX as explicit controls.
Level vocabulary (= our method)
OTE pullback (1st choice) · hold-&-retest (2nd) · no-chase if extended without retest · wick ≠ failure (acceptance is what counts) · invalidation = acceptance beyond the risk floor.
Where we stand vs Talon (honest gap)
- ✅ Architecture: universe scan → strict/watch → score → index throttle → timing → invalidation = our spec §0.2/§5/§6/§6.2. Validated from the outside.
- ✅ Index throttle: v_joint (proven Sharpe 1.14) + the touchstone.
- ✅ Families/themes: families.csv (684 tickers, 30 families) + rotation §6.2.
- 🔴 THE GAP = per-strike VEX/GEX ladders (Best Entry/OTE, VEX ladder, structural invalidation levels). That comes from the per-strike GEX/VEX map = UW / Skylit per-strike data (post-Friday). Without it, we have the architecture but not the levels.
- 🔴 Weekly presentation layer: generate this report (sections + per-name + hedge dashboard + rotation) from the
Setupschema → to code (V1 output). - ⚠️ Rotation engine: Talon has a formal "rotation confirmed / none" detector. Our §6.2 is descriptive → to gauntlet.
Verdict
~80% of the architecture is specified/built; the missing 20% = per-strike data (UW) + the weekly presentation layer. Talon proves the target format is exactly our V1. It is the reference output spec: the dev builds toward THIS format.