Apply fully-invested risk lower bound only to fully-invested portfolios

pypfopt/efficient_frontier/efficient_frontier.py

@@ -386,14 +386,14 @@ def efficient_risk(self, target_volatility, market_neutral=False):
         if not isinstance(target_volatility, (float, int)) or target_volatility < 0:
             raise ValueError("target_volatility should be a positive float")
 
-        global_min_volatility = np.sqrt(1 / np.sum(np.linalg.pinv(self.cov_matrix)))
-
-        if target_volatility < global_min_volatility:
-            raise ValueError(
-                "The minimum volatility is {:.3f}. Please use a higher target_volatility".format(
-                    global_min_volatility
+        if not market_neutral:
+            global_min_volatility = np.sqrt(1 / np.sum(np.linalg.pinv(self.cov_matrix)))
+            if target_volatility < global_min_volatility:
+                raise ValueError(
+                    "The minimum volatility is {:.3f}. Please use a higher target_volatility".format(
+                        global_min_volatility
+                    )
                 )
-            )
 
         update_existing_parameter = self.is_parameter_defined("target_variance")
         if update_existing_parameter:

tests/test_market_neutral_risk_regression.py

@@ -0,0 +1,37 @@
+import numpy as np
+import pytest
+
+from pypfopt import EfficientFrontier, exceptions
+
+
+@pytest.mark.parametrize("target", [0.01, 0.05, 0.1])
+def test_market_neutral_risk_below_fully_invested_minimum(target):
+    # With diagonal 0.04 covariance and weights (a, -a), risk=sqrt(0.08)*a.
+    ef = EfficientFrontier([0.1, 0.0], np.eye(2) * 0.04, weight_bounds=(-1, 1))
+    weights = np.array(list(ef.efficient_risk(target, market_neutral=True).values()))
+    a = target / np.sqrt(0.08)
+    np.testing.assert_allclose(weights, [a, -a], atol=2e-5)
+
+
+def test_market_neutral_risk_updates_parameter():
+    ef = EfficientFrontier([0.1, 0.0], np.eye(2) * 0.04, weight_bounds=(-1, 1))
+    for target in (0.05, 0.02):
+        weights = np.array(
+            list(ef.efficient_risk(target, market_neutral=True).values())
+        )
+        np.testing.assert_allclose(
+            weights, [target / np.sqrt(0.08), -target / np.sqrt(0.08)], atol=2e-5
+        )
+
+
+def test_fully_invested_low_risk_is_still_rejected():
+    ef = EfficientFrontier([0.1, 0.0], np.eye(2) * 0.04)
+    with pytest.raises(ValueError, match="minimum volatility"):
+        ef.efficient_risk(0.05)
+
+
+def test_market_neutral_infeasible_extra_constraint_is_not_accepted():
+    ef = EfficientFrontier([0.1, 0.0], np.eye(2) * 0.04, weight_bounds=(-1, 1))
+    ef.add_constraint(lambda w: w[0] >= 0.5)
+    with pytest.raises(exceptions.OptimizationError):
+        ef.efficient_risk(0.05, market_neutral=True)