def test_olmar_alpha1_error(self):
"""
Tests ValueError if reversion method is 2 and alpha is less than 1.
"""
# Initialize OLMAR.
olmar5 = OLMAR(reversion_method=2, epsilon=2, alpha=-1)
with self.assertRaises(ValueError):
# Running allocate will raise ValueError.
olmar5.allocate(self.data)
def test_olmar_window_error(self):
"""
Tests ValueError if reversion method is 1 and window is less than 1.
"""
# Initialize OLMAR.
olmar3 = OLMAR(reversion_method=1, epsilon=2, window=0)
with self.assertRaises(ValueError):
# Running allocate will raise ValueError.
olmar3.allocate(self.data)
def test_olmar_epsilon_error(self):
"""
Tests ValueError if epsilon is below than 1.
"""
# Initialize OLMAR.
olmar2 = OLMAR(reversion_method=1, epsilon=0, window=10)
with self.assertRaises(ValueError):
# Running allocate will raise ValueError.
olmar2.allocate(self.data)
def test_olmar_edge_case_error(self):
"""
Tests that lambd returns 0 if predicted change is mean change.
"""
# Initialize OLMAR.
olmar7 = OLMAR(reversion_method=1, epsilon=2, window=1)
no_change_data = self.data
no_change_data.iloc[:] = 1
olmar7.allocate(no_change_data, resample_by='M')
def test_olmar1_solution(self):
"""
Test the calculation of online moving average reversion with the second reversion method.
"""
# Initialize OLMAR.
olmar1 = OLMAR(reversion_method=2, epsilon=10, alpha=0.5)
# Allocates asset prices to OLMAR.
olmar1.allocate(self.data, resample_by='M')
# Create np.array of all_weights.
all_weights = np.array(olmar1.all_weights)
# Check if all weights sum to 1.
for i in range(all_weights.shape[0]):
weights = all_weights[i]
assert (weights >= 0).all()
assert len(weights) == self.data.shape[1]
np.testing.assert_almost_equal(np.sum(weights), 1)