def test_get_annualized_mean(data_cube, frequency):
expected = [
-1.58544394e-04, 3.95923527e-05, 8.10572621e-04, -3.15037014e-03,
-8.93454642e-04
]
assert_almost_equal(get_annualized_mean(data_cube, frequency), expected)
assert_almost_equal(
[get_annualized_mean(data_cube[..., i], frequency) for i in range(N)],
expected)
def mean_best_guess():
space = 2**np.linspace(-15, 4, 30)
space = np.sort([*-space, *space])
f_space = np.array([
get_annualized_mean(self.data + x, self.time_unit) - mean
for x in space
])
mask: np.ndarray = (f_space[:-1] * f_space[1:]) <= 0
return get_by_bisection(space, f_space,
mask) if any(mask) else get_closest_to_0(
space, f_space)
def statistics(self):
"""
Returns the statistics (4 moments) of the cube
Returns
-------
DataFrame
The first 4 moments of the cube for each asset (last axis)
"""
return pd.DataFrame({
"Mean":
get_annualized_mean(self, self.time_unit),
"SD":
get_annualized_sd(self, self.time_unit),
"Skew":
get_annualized_skew(self, self.time_unit),
"Kurt":
get_annualized_kurtosis(self, self.time_unit),
})
def annualized_moments(self, x: np.ndarray, mean: float, sd: float):
calibrated_data = self.data * x[1] + x[0]
return (get_annualized_mean(calibrated_data, self.time_unit) - mean,
get_annualized_sd(calibrated_data, self.time_unit) - sd)
def annualized_mean(self, x: float, target: float):
return get_annualized_mean(self.data + x, self.time_unit) - target