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/recommendation_service.py

@@ -44,6 +44,16 @@ _MODERATE_MAX_ATR = 4.6
 # the same tolerance the chart and alerts use, so S/R is one model app-wide.
 _SR_ZONE_TOLERANCE = 0.02
 
+# Measured-move projection used when a ticker has NO S/R level overhead in the
+# trade direction (genuine blue-sky, e.g. a stock at all-time highs). Without
+# this the scanner produces no setup and the last (now stale) one lingers. The
+# projected target sits this many ATRs from entry, so with the default 1.5-ATR
+# stop it is a clean 2:1 R:R. Projected targets carry no touch history, so they
+# take near-zero strength (small probability haircut via the strength magnet) and
+# face a stricter activation bar — see app/services/qualification.py.
+PROJECTED_TARGET_ATR_MULTIPLE = 3.0
+PROJECTED_TARGET_STRENGTH = 10.0
+
 
 def _clamp(value: float, low: float, high: float) -> float:
     return max(low, min(high, value))
@@ -310,7 +320,43 @@ class TargetGenerator:
             )
 
         if not candidates:
-            return []
+            # No S/R level in the trade direction cleared the ATR distance
+            # filter. If there is genuinely NO S/R overhead at all (blue-sky,
+            # e.g. all-time highs), project a measured-move target so a breakout
+            # name still yields a setup. When overhead S/R DOES exist but was
+            # merely too close/far to qualify, produce nothing as before — we
+            # never project a target through real, nearby resistance.
+            #
+            # Check both the level's tag AND its price. Zone representatives are
+            # typed relative to entry, so a resistance cluster straddling entry
+            # counts as overhead even if its near edge sits just below (which
+            # keeps this aligned with the scanner's raw ``levels_above`` gate);
+            # the price comparison covers raw levels for other callers.
+            has_overhead = any(
+                (direction == "long" and (lv.type == "resistance" or lv.price_level > entry_price))
+                or (direction == "short" and (lv.type == "support" or lv.price_level < entry_price))
+                for lv in sr_levels
+            )
+            if has_overhead:
+                return []
+            projected_price = (
+                entry_price + PROJECTED_TARGET_ATR_MULTIPLE * atr_value
+                if direction == "long"
+                else entry_price - PROJECTED_TARGET_ATR_MULTIPLE * atr_value
+            )
+            reward = abs(projected_price - entry_price)
+            return [
+                {
+                    "price": float(projected_price),
+                    "distance_from_entry": float(reward),
+                    "distance_atr_multiple": float(reward / atr_value),
+                    "rr_ratio": float(reward / risk),
+                    "classification": "Moderate",
+                    "sr_level_id": -1,
+                    "sr_strength": float(PROJECTED_TARGET_STRENGTH),
+                    "projected": True,
+                }
+            ]
 
         # Select up to 5 targets that SPAN the distance range, instead of the
         # top-5 by quality (which biases toward far, high-R:R levels and buries
@@ -450,9 +496,11 @@ def _choose_recommended_action(
     """Pick the ticker action — but only recommend a direction you can trade.
 
     A direction is recommendable only if a tradeable setup exists for it
-    (``available_directions``). So a strong LONG bias on a stock at all-time
-    highs — where the scanner can build no long target — does NOT yield
-    LONG_HIGH; it falls through to NEUTRAL, and the reasoning explains why.
+    (``available_directions``). A strong LONG bias on a stock with no tradeable
+    long setup does NOT yield LONG_HIGH; it falls through to NEUTRAL, and the
+    reasoning explains why. (At genuine all-time highs the scanner now projects a
+    measured-move long target, so blue-sky names can be recommendable; a name
+    capped just under resistance — with no ≥threshold R:R — still cannot.)
     """
     high = float(config.get("recommendation_high_confidence_threshold", 70.0))
     moderate = float(config.get("recommendation_moderate_confidence_threshold", 50.0))
@@ -658,7 +706,14 @@ async def enhance_trade_setup(
 
     # Per-setup conflicts (target availability is specific to this setup)
     setup_conflicts = list(conflicts)
-    if len(targets) < 3:
+    primary_projected = bool(primary is not None and primary.get("projected"))
+    if primary_projected:
+        # Blue-sky: no overhead S/R to anchor to. Flag it so the target's basis
+        # is explicit rather than looking like a normal S/R level.
+        setup_conflicts.append(
+            "projected-target: No overhead resistance — target is an ATR measured-move projection"
+        )
+    elif len(targets) < 3:
         setup_conflicts.append("target-availability: Fewer than 3 valid S/R targets available")
 
     # Action and reasoning are ticker-level: they consider both directions and