Fix scoring/recommendation correctness and calibration

Triggered by CNC showing "LONG (High Confidence)" with SHORT reasoning
and no long setup.

- A: recommendation action + reasoning are ticker-level and identical
  on both setups; reasoning always matches the shown action
- B: recommended_action only picks a direction with a tradeable setup;
  strong bias with no setup (e.g. price at ATH) → NEUTRAL with an
  explanatory reason instead of a fake LONG_HIGH
- C: confidence is a directional-agreement model — opposing signals push
  it below 50 (SHORT on a 92-technical/99-momentum stock ~0%, not 55%)
- D: fundamental score requires >=2 real metrics (market-cap-only no
  longer yields a high score)
- E: RSI score peaks at healthy momentum (~60) and penalizes
  overbought/oversold extremes instead of treating RSI 90 as maximal
- F: fundamentals chain merges fields across providers (FMP market cap
  + Finnhub P/E) instead of stopping at the first with any field
- NEUTRAL label: "No Clear Setup" (covers untradeable-bias case)

Scores recompute on next scan/scoring run; C and E shift score
distributions intentionally.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

tests/unit/test_recommendation_service.py

@@ -3,12 +3,20 @@ from __future__ import annotations
 from dataclasses import dataclass
 
 from app.services.recommendation_service import (
+    _build_reasoning,
+    _choose_recommended_action,
     direction_analyzer,
     probability_estimator,
     signal_conflict_detector,
     target_generator,
 )
 
+_DEFAULT_CFG = {
+    "recommendation_high_confidence_threshold": 70.0,
+    "recommendation_moderate_confidence_threshold": 50.0,
+    "recommendation_confidence_diff_threshold": 20.0,
+}
+
 
 @dataclass
 class _SRLevelStub:
@@ -52,6 +60,48 @@ def test_high_confidence_short_example():
     assert confidence > 70.0
 
 
+def test_short_confidence_low_on_strongly_bullish_stock():
+    """The CNC case: technical 92 / momentum 99 must make SHORT confidence low."""
+    dims = {"technical": 92.0, "momentum": 99.0, "fundamental": 96.0}
+
+    short_conf = direction_analyzer.calculate_confidence(
+        direction="short", dimension_scores=dims, sentiment_classification="neutral", conflicts=[],
+    )
+    long_conf = direction_analyzer.calculate_confidence(
+        direction="long", dimension_scores=dims, sentiment_classification="neutral", conflicts=[],
+    )
+
+    assert short_conf < 20.0   # not 55
+    assert long_conf > 90.0
+
+
+def test_action_neutral_when_bias_direction_has_no_setup():
+    """Strong LONG bias but only a SHORT setup is tradeable → NEUTRAL, not LONG_HIGH."""
+    action = _choose_recommended_action(
+        long_confidence=100.0, short_confidence=5.0, config=_DEFAULT_CFG,
+        available_directions={"short"},
+    )
+    assert action == "NEUTRAL"
+
+    # With the long setup available, the same numbers give LONG_HIGH
+    action_ok = _choose_recommended_action(
+        long_confidence=100.0, short_confidence=5.0, config=_DEFAULT_CFG,
+        available_directions={"long", "short"},
+    )
+    assert action_ok == "LONG_HIGH"
+
+
+def test_reasoning_explains_missing_setup():
+    reasoning = _build_reasoning(
+        action="NEUTRAL", long_confidence=100.0, short_confidence=5.0, conflicts=[],
+        dimension_scores={"technical": 92.0, "momentum": 99.0},
+        sentiment_classification="neutral", config=_DEFAULT_CFG,
+        available_directions={"short"},
+    )
+    assert "bias is LONG" in reasoning
+    assert "no high-conviction long setup" in reasoning.lower()
+
+
 def test_detects_sentiment_technical_conflict():
     conflicts = signal_conflict_detector.detect_conflicts(
         dimension_scores={"technical": 72.0, "momentum": 55.0, "fundamental": 50.0},