replace EV activation gate with cross-sectional 12-1 momentum ranking

The 5-year backtest confirmed the EV gate adds negative value (high threshold =
worst expectancy) and that 12-1 month momentum is the one price signal with a
plausible, right-signed cross-sectional IC (~0.05). So "qualified" now means:
clears the R:R + confidence floors AND the ticker ranks in the top
`min_momentum_percentile` of the universe by 12-1 momentum that week.

- qualification.py: drop expected_value_r / the EV gate; add a momentum-percentile
  gate (duck-typed `momentum_percentile`, only enforced when attached + threshold
  set, else defers to floors). Mirrored in frontend qualification.ts.
- activation config/schema: min_expected_value -> min_momentum_percentile
  (default 80 = top quintile). ActivationSettings, DashboardPage (ranks/【shows】
  momentum instead of EV), and the BacktestPanel sweep follow.
- backtest: rank each ISO week's universe by 12-1 momentum, assign a percentile,
  and qualify the top slice; the sweep now sweeps the percentile cutoff.

Also offload the backtest's per-ticker compute to a worker thread so the heavy
~5y run no longer blocks the API event loop (the "backend offline" flicker).

Production setups don't carry momentum_percentile yet — wiring the scanner to
attach it (a universe momentum-rank step) is the next step; until then the live
gate defers to floors while the backtest measures the momentum selection. 330
backend tests pass; frontend build clean.

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

tests/unit/test_activation_settings.py

@@ -25,7 +25,7 @@ class TestActivationConfig:
     async def test_defaults_when_unset(self, session: AsyncSession):
         config = await get_activation_config(session)
         assert config == {
-            "min_expected_value": 0.15,
+            "min_momentum_percentile": 80.0,
             "min_rr": 1.2,
             "min_confidence": 55.0,
             "min_target_probability": 0.0,
@@ -35,13 +35,13 @@ class TestActivationConfig:
 
     async def test_update_and_read_back(self, session: AsyncSession):
         updated = await update_activation_config(
-            session, {"min_expected_value": 0.25, "min_confidence": 60.0}
+            session, {"min_momentum_percentile": 70.0, "min_confidence": 60.0}
         )
-        assert updated["min_expected_value"] == 0.25
+        assert updated["min_momentum_percentile"] == 70.0
         assert updated["min_confidence"] == 60.0
 
         config = await get_activation_config(session)
-        assert config["min_expected_value"] == 0.25
+        assert config["min_momentum_percentile"] == 70.0
         assert config["min_confidence"] == 60.0
 
     async def test_partial_update_keeps_other_value(self, session: AsyncSession):
@@ -50,9 +50,9 @@ class TestActivationConfig:
         assert config["min_rr"] == 1.2  # default untouched
         assert config["min_confidence"] == 80.0
 
-    async def test_rejects_out_of_range_expected_value(self, session: AsyncSession):
+    async def test_rejects_out_of_range_momentum_percentile(self, session: AsyncSession):
         with pytest.raises(ValidationError):
-            await update_activation_config(session, {"min_expected_value": 50.0})
+            await update_activation_config(session, {"min_momentum_percentile": 150.0})
 
     async def test_conviction_flags_round_trip(self, session: AsyncSession):
         await update_activation_config(

tests/unit/test_backtest_service.py

@@ -113,8 +113,8 @@ async def test_run_backtest_smoke(session):
     # the oscillating series should yield at least some resolved setups
     assert report["candidates"] >= 1
 
-    # sweep: lowering the EV threshold can only add qualifiers, never remove them
-    sweep = sorted(report["sweep"], key=lambda r: r["min_expected_value"], reverse=True)
+    # sweep: lowering the momentum-percentile cutoff can only add qualifiers
+    sweep = sorted(report["sweep"], key=lambda r: r["min_momentum_percentile"], reverse=True)
     counts = [r["total"] for r in sweep]
     assert counts == sorted(counts)  # ascending as threshold descends
     # every calibration row is internally consistent

tests/unit/test_qualification.py

@@ -1,20 +1,15 @@
-"""Unit tests for the activation qualification predicate (EV-based gate)."""
+"""Unit tests for the activation qualification predicate (momentum-based gate)."""
 
 from __future__ import annotations
 
 from types import SimpleNamespace
 
-from app.services.qualification import (
-    best_target_probability,
-    expected_value_r,
-    primary_target_probability,
-    setup_qualifies,
-)
+from app.services.qualification import best_target_probability, setup_qualifies
 
-# Default gate: expected value is the core test; conviction/conflict/target-prob
-# are optional tighteners, off here.
+# Default gate: floors only; the momentum selection is off (0). Conviction /
+# conflict / target-probability are optional tighteners, off here.
 DEFAULT_GATE = {
-    "min_expected_value": 0.15,
+    "min_momentum_percentile": 0.0,
     "min_rr": 1.2,
     "min_confidence": 55.0,
     "min_target_probability": 0.0,
@@ -22,9 +17,12 @@ DEFAULT_GATE = {
     "exclude_conflicts": False,
 }
 
-# Strict gate: every optional tightener turned on (the old shipped defaults).
+# Gate with the cross-sectional momentum selection on (top quintile).
+MOMENTUM_GATE = {**DEFAULT_GATE, "min_momentum_percentile": 80.0}
+
+# Strict gate: every optional tightener turned on.
 STRICT_GATE = {
-    "min_expected_value": 0.0,
+    "min_momentum_percentile": 0.0,
     "min_rr": 2.0,
     "min_confidence": 70.0,
     "min_target_probability": 60.0,
@@ -45,59 +43,17 @@ def _setup(**kwargs):
     return SimpleNamespace(**base)
 
 
-class TestExpectedValue:
-    def test_uses_primary_target_not_best(self):
-        s = _setup(
-            rr_ratio=1.5,
-            targets=[
-                {"probability": 80.0},
-                {"probability": 30.0, "is_primary": True},
-            ],
-        )
-        # EV from the primary (30%): 0.3*1.5 - 0.7 = -0.25
-        assert expected_value_r(s) == -0.25
-        assert primary_target_probability(s) == 30.0
-
-    def test_falls_back_to_best_when_no_primary_flag(self):
-        s = _setup(rr_ratio=2.0, targets=[{"probability": 40.0}, {"probability": 60.0}])
-        assert primary_target_probability(s) == 60.0
-        # 0.6*2.0 - 0.4 = 0.8
-        assert abs(expected_value_r(s) - 0.8) < 1e-9
-
-    def test_none_when_no_targets(self):
-        assert expected_value_r(_setup(targets=[])) is None
-        assert primary_target_probability(_setup(targets=[])) is None
-
-
-class TestSetupQualifies:
-    def test_positive_ev_setup_passes(self):
-        # primary 50% @ rr 3.0 → EV = 1.0
+class TestFloors:
+    def test_passes_floors(self):
         assert setup_qualifies(_setup(), DEFAULT_GATE) is True
 
-    def test_negative_ev_fails(self):
-        # primary 30% @ rr 1.3 → EV = -0.31, below the 0.15 floor
-        s = _setup(rr_ratio=1.3, targets=[{"probability": 30.0, "is_primary": True}])
-        assert setup_qualifies(s, DEFAULT_GATE) is False
-
-    def test_thin_positive_ev_below_floor_fails(self):
-        # Positive but thin: 0.45*1.3 - 0.55 = 0.035, under the 0.15 floor.
-        s = _setup(rr_ratio=1.3, targets=[{"probability": 45.0, "is_primary": True}])
-        assert setup_qualifies(s, DEFAULT_GATE) is False
-
     def test_low_rr_floor_fails(self):
         assert setup_qualifies(_setup(rr_ratio=1.0), DEFAULT_GATE) is False
 
     def test_low_confidence_fails(self):
         assert setup_qualifies(_setup(confidence_score=40.0), DEFAULT_GATE) is False
 
-    def test_no_targets_defers_to_rr_and_confidence(self):
-        # No probability → EV uncomputable → not blocked on EV; passes on floors.
-        assert setup_qualifies(_setup(targets=[]), DEFAULT_GATE) is True
-        # ...but still subject to the rr/confidence floors.
-        assert setup_qualifies(_setup(targets=[], rr_ratio=1.0), DEFAULT_GATE) is False
-
     def test_conviction_and_conflict_ignored_by_default(self):
-        # Moderate action + medium risk still pass when tighteners are off.
         s = _setup(recommended_action="LONG_MODERATE", risk_level="Medium")
         assert setup_qualifies(s, DEFAULT_GATE) is True
 
@@ -113,11 +69,26 @@ class TestSetupQualifies:
         s = _setup(direction="long", target=120.0, stop_loss=95.0, current_price=94.0)
         assert setup_qualifies(s, DEFAULT_GATE) is False
 
-    def test_missing_min_ev_key_skips_ev(self):
-        # Legacy callers without min_expected_value: EV defaults to -inf (no floor).
-        legacy = {k: v for k, v in DEFAULT_GATE.items() if k != "min_expected_value"}
-        s = _setup(rr_ratio=1.3, targets=[{"probability": 30.0, "is_primary": True}])
-        assert setup_qualifies(s, legacy) is True
+
+class TestMomentumGate:
+    def test_top_momentum_passes(self):
+        assert setup_qualifies(_setup(momentum_percentile=92.0), MOMENTUM_GATE) is True
+
+    def test_below_threshold_fails(self):
+        assert setup_qualifies(_setup(momentum_percentile=50.0), MOMENTUM_GATE) is False
+
+    def test_missing_percentile_defers_to_floors(self):
+        # No percentile attached (e.g. production not yet wired) → the momentum
+        # gate is skipped and the setup still clears on the floors.
+        assert setup_qualifies(_setup(), MOMENTUM_GATE) is True
+
+    def test_threshold_zero_disables_gate(self):
+        # min_momentum_percentile 0 → a low-momentum name still passes.
+        assert setup_qualifies(_setup(momentum_percentile=10.0), DEFAULT_GATE) is True
+
+    def test_missing_key_defaults_off(self):
+        legacy = {k: v for k, v in DEFAULT_GATE.items() if k != "min_momentum_percentile"}
+        assert setup_qualifies(_setup(momentum_percentile=10.0), legacy) is True
 
 
 class TestStrictTighteners: