Add blue-sky projected targets and played-out setup UX

Fixes stale below-price setups showing as current recommendations. Three
distinct causes share the symptom (get_trade_setups returns the latest stored
setup per direction and never expires it):

- Genuine blue-sky (no overhead S/R): scanner + TargetGenerator now project a
  measured-move target (entry +/- 3*ATR, ~2:1 R:R), flagged projected with a
  low sr_strength probability haircut. Overhead check keys on level tag OR price
  so it never projects through a straddling resistance cluster.
- Projected targets clear a stricter activation bar (long-only, momentum >= 90,
  confidence >= min+10), independent of the general momentum gate. Mirrored in
  frontend qualification.ts.
- Played-out UX (fixes the reported TTWO case, which is R:R-starved under a
  resistance cluster, not blue-sky): when price is at/past target or through the
  stop, RecommendationPanel shows a "No current setup" state and softens the
  stale ticker-level header/reasoning, instead of a stale actionable card.

No migration: the projected flag rides in existing targets_json. 504 backend
unit tests pass; frontend typechecks.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

app/services/qualification.py

@@ -16,6 +16,15 @@ from typing import Any
 
 HIGH_CONVICTION_ACTIONS = {"LONG_HIGH", "SHORT_HIGH"}
 
+# A projected (blue-sky) target has no S/R validation — it is a measured-move
+# extension used when nothing sits overhead. Because that is exactly the kind of
+# unvalidated target the gate exists to distrust, a projected setup clears a
+# STRICTER bar than an S/R-anchored one, regardless of whether the general
+# momentum gate is active: long-only (breakout continuation), strong residual
+# momentum, and a higher confidence floor. Mirrored in frontend/src/lib/qualification.ts.
+PROJECTED_MIN_MOMENTUM_PERCENTILE = 90.0
+PROJECTED_CONFIDENCE_MARGIN = 10.0
+
 
 def _action_direction(action: str | None) -> str:
     if not action or action == "NEUTRAL":
@@ -46,6 +55,19 @@ def primary_target_probability(setup: Any) -> float | None:
     return best if best > 0 else None
 
 
+def primary_target_is_projected(setup: Any) -> bool:
+    """Whether the setup's headline target is a blue-sky measured-move projection.
+
+    Prefers the starred primary; falls back to any projected target when none is
+    explicitly flagged primary (matches primary_target_probability's fallback).
+    """
+    targets = getattr(setup, "targets", None) or []
+    for target in targets:
+        if isinstance(target, dict) and target.get("is_primary"):
+            return bool(target.get("projected"))
+    return any(isinstance(t, dict) and t.get("projected") for t in targets)
+
+
 def live_risk_reward(setup: Any, current_price: float) -> float | None:
     """R:R recomputed from the CURRENT price, not the (possibly stale) entry.
 
@@ -102,6 +124,18 @@ def setup_qualifies(setup: Any, config: dict) -> bool:
         momentum_percentile = getattr(setup, "momentum_percentile", None)
         if momentum_percentile is None or momentum_percentile < min_pct:
             return False
+    # Projected (blue-sky) targets clear a stricter bar than S/R-anchored ones,
+    # independent of the general momentum gate above: long-only, strong residual
+    # momentum, and a higher confidence floor. The target has no S/R validation,
+    # so we only trust it for high-momentum breakout continuations.
+    if primary_target_is_projected(setup):
+        if (getattr(setup, "direction", "long") or "long").lower() != "long":
+            return False
+        momentum_percentile = getattr(setup, "momentum_percentile", None)
+        if momentum_percentile is None or momentum_percentile < PROJECTED_MIN_MOMENTUM_PERCENTILE:
+            return False
+        if (setup.confidence_score or 0.0) < config["min_confidence"] + PROJECTED_CONFIDENCE_MARGIN:
+            return False
     # A setup is actionable only when the live ticker action points in the same
     # direction. NEUTRAL means no clear signal; an opposite action means the
     # setup is counter-bias. ``exclude_neutral`` defaults on; callers that omit