ENH: Add explicit error for bad horizon

Check that horizon is an int >= 1

arch/tests/unitroot/test_engle_granger.py

@@ -171,6 +171,8 @@ def test_trivariate(data, trend, method, lhs):
 
         fig = test.plot()
         assert isinstance(fig, plt.Figure)
+
+        plt.close("all")
     assert isinstance(test.resid, pd.Series)
     assert test.resid.shape[0] == dep.shape[0]
 

arch/tests/univariate/test_forecast.py

@@ -123,6 +123,8 @@ def test_zero_mean_forecast(self):
         assert_allclose(
             fcast.residual_variance.iloc[499:], np.ones((501, 3)) * params[0]
         )
+        with pytest.raises(ValueError, match="horizon must be an integer >= 1"):
+            res.forecast(horizon=0)
 
     def test_frame_labels(self):
         am = arch_model(self.zero_mean, mean="Zero", vol="Constant")

arch/tests/univariate/test_mean.py

@@ -132,6 +132,8 @@ def test_constant_mean(self):
         assert isinstance(cm.__repr__(), str)
         assert isinstance(cm.__str__(), str)
         assert "<strong>" in cm._repr_html_()
+        with pytest.raises(ValueError, match="horizon must be an integer >= 1"):
+            res.forecast(horizon=0, start=20)
 
     def test_zero_mean(self):
         zm = ZeroMean(self.y)
@@ -456,13 +458,19 @@ def test_ar_plot(self):
         with pytest.raises(ValueError):
             res.plot(annualize="unknown")
 
+        import matplotlib.pyplot as plt
+
+        plt.close("all")
+
         res.plot(scale=360)
         res.hedgehog_plot(start=500)
         res.hedgehog_plot(start=500, plot_type="mean")
         res.hedgehog_plot(plot_type="volatility")
         res.hedgehog_plot(start=500, method="simulation", simulations=100)
         res.hedgehog_plot(plot_type="volatility", method="bootstrap")
 
+        plt.close("all")
+
     def test_arch_arx(self):
         self.rng.seed(12345)
         x = self.rng.randn(500, 3)

arch/tests/univariate/test_variance_forecasting.py

@@ -235,6 +235,8 @@ def test_arch_1_forecast(self):
         )
         expected[:100] = np.nan
         assert_allclose(forecast.forecasts, expected)
+        with pytest.raises(ValueError, match="horizon must be an integer >= 1"):
+            vol.forecast(params, self.resid, backcast, var_bounds, start=0, horizon=0)
 
     def test_arch_1_forecast_simulation(self):
         dist = Normal(self.rng)

arch/unitroot/critical_values/simulation/adf_z_critical_values_simulation_joblib.py

@@ -6,10 +6,8 @@
 import random
 
 import colorama
-from joblib import Parallel, delayed
 import numpy as np
 from numpy.random import PCG64, Generator, SeedSequence
-import psutil
 
 from adf_simulation import (
     OUTPUT_PATH,
@@ -18,6 +16,8 @@
     TRENDS,
     adf_simulation,
 )
+from joblib import Parallel, delayed
+import psutil
 
 GREEN = colorama.Fore.GREEN
 BLUE = colorama.Fore.BLUE

arch/unitroot/critical_values/simulation/engle_granger_simulation.py

@@ -11,10 +11,11 @@
 from typing import List
 
 import colorama
-from joblib import Parallel, cpu_count, delayed
 import numpy as np
 from numpy.random import PCG64, Generator, SeedSequence
 
+from joblib import Parallel, cpu_count, delayed
+
 ROOT = os.path.join(os.path.split(os.path.abspath(__file__))[0], "engle-granger")
 if not os.path.exists(ROOT):
     os.mkdir(ROOT)

arch/unitroot/critical_values/simulation/phillips-ouliaris-simulation.py

@@ -8,15 +8,15 @@
 from typing import Optional, Tuple
 
 import colorama
-from joblib import Parallel, delayed
 import numpy as np
 from numpy.linalg import inv, lstsq, solve
 import pandas as pd
-import psutil
 
 from arch.typing import NDArray
 from arch.utility.timeseries import add_trend
+from joblib import Parallel, delayed
 from phillips_ouliaris import QUANTILES, ROOT, SAMPLE_SIZES, TRENDS
+import psutil
 
 GREEN = colorama.Fore.GREEN
 BLUE = colorama.Fore.BLUE

arch/unitroot/unitroot.py

@@ -1341,9 +1341,9 @@ def _autolag(self) -> None:
             s1 += i * resids_prod
         if s0 <= 0:
             raise InfeasibleTestException(
-                f"Residuals are all zero and so automatic bandwidth selection cannot "
-                f"be used. This is usually an indication that the series being testes "
-                f"is too small or have constant values."
+                "Residuals are all zero and so automatic bandwidth selection cannot "
+                "be used. This is usually an indication that the series being testes "
+                "is too small or have constant values."
             )
         s_hat = s1 / s0
         pwr = 1.0 / 3.0

arch/univariate/base.py

@@ -1388,6 +1388,8 @@ def forecast(
                 or params.ndim != self._params.ndim
             ):
                 raise ValueError("params have incorrect dimensions")
+        if not isinstance(horizon, (int, np.integer)) or horizon < 1:
+            raise ValueError("horizon must be an integer >= 1.")
         return self.model.forecast(
             params, horizon, start, align, method, simulations, rng, random_state
         )

arch/univariate/mean.py

@@ -692,6 +692,8 @@ def forecast(
         rng: Optional[Callable[[Union[int, Tuple[int, ...]]], NDArray]] = None,
         random_state: Optional[np.random.RandomState] = None,
     ) -> ARCHModelForecast:
+        if not isinstance(horizon, (int, np.integer)) or horizon < 1:
+            raise ValueError("horizon must be an integer >= 1.")
         # Check start
         earliest, default_start = self._fit_indices
         default_start = max(0, default_start - 1)

arch/univariate/volatility.py

@@ -643,7 +643,8 @@ def forecast(
         method = method.lower()
         if method not in ("analytic", "simulation", "bootstrap"):
             raise ValueError("{0} is not a known forecasting method".format(method))
-
+        if not isinstance(horizon, (int, np.integer)) or horizon < 1:
+            raise ValueError("horizon must be an integer >= 1.")
         self._check_forecasting_method(method, horizon)
         start = len(resids) - 1 if start is None else start
         if method == "analytic":
@@ -2743,7 +2744,7 @@ def backcast(self, resids: NDArray) -> float:
 
     def bounds(self, resids: NDArray) -> List[Tuple[float, float]]:
         if not self._unit_scale:
-            v = self.starting_values(resids)
+            v = float(np.squeeze(self.starting_values(resids)))
             _resids = resids / np.sqrt(self._variance[self._start : self._stop])
             mu = _resids.mean()
             return [(v / 100000.0, 10.0 * (v + mu ** 2.0))]

setup.cfg

@@ -29,6 +29,7 @@ filterwarnings =
     error:invalid value encountered in log:RuntimeWarning
     error:divide by zero encountered in double_scalars:RuntimeWarning
     error:Lag selection has changed:DeprecationWarning
+    error:More than 20 figures have been opened:RuntimeWarning:
 markers =
     slow: mark a test as slow