def test_train_valid_test_split(fn):
columns = ['test column 1', 'test column 2', 'test column 3']
dates = generate_random_dates(10)
assets = get_assets(3)
index = pd.MultiIndex.from_product([dates, assets])
values = np.arange(len(index) * len(columns)).reshape(
[len(columns), len(index)]).T
targets = np.arange(len(index))
fn_inputs = {
'all_x': pd.DataFrame(values, index, columns),
'all_y': pd.Series(targets, index, name='target'),
'train_size': 0.6,
'valid_size': 0.2,
'test_size': 0.2
}
fn_correct_outputs = OrderedDict([
('X_train', pd.DataFrame(values[:18], index[:18], columns=columns)),
('X_valid', pd.DataFrame(values[18:24], index[18:24],
columns=columns)),
('X_test', pd.DataFrame(values[24:], index[24:], columns=columns)),
('y_train', pd.Series(targets[:18], index[:18])),
('y_valid', pd.Series(targets[18:24], index[18:24])),
('y_test', pd.Series(targets[24:], index[24:]))
])
assert_output(fn,
fn_inputs,
fn_correct_outputs,
check_parameter_changes=False)
def test_non_overlapping_samples(fn):
columns = ['test column 1', 'test column 2']
dates = generate_random_dates(8)
assets = get_assets(3)
index = pd.MultiIndex.from_product([dates, assets])
values = np.arange(len(index) * len(columns)).reshape(
[len(columns), len(index)]).T
targets = np.arange(len(index))
fn_inputs = {
'x': pd.DataFrame(values, index, columns),
'y': pd.Series(targets, index),
'n_skip_samples': 2,
'start_i': 1
}
new_index = pd.MultiIndex.from_product(
[dates[fn_inputs['start_i']::fn_inputs['n_skip_samples'] + 1], assets])
fn_correct_outputs = OrderedDict([
('non_overlapping_x',
pd.DataFrame([[3, 27], [4, 28], [5, 29], [12, 36], [13, 37], [14, 38],
[21, 45], [22, 46], [23, 47]], new_index, columns)),
('non_overlapping_y',
pd.Series([3, 4, 5, 12, 13, 14, 21, 22, 23], new_index))
])
assert_output(fn,
fn_inputs,
fn_correct_outputs,
check_parameter_changes=False)
def test_get_long_short(fn):
tickers = generate_random_tickers(3)
dates = generate_random_dates(4)
fn_inputs = {
'close':
pd.DataFrame([[25.6788, 35.1392, 34.0527], [25.1884, 14.3453, 39.9373],
[78.2803, 34.3854, 23.2932], [88.8725, 52.223, 34.4107]],
dates, tickers),
'lookback_high':
pd.DataFrame(
[[np.nan, np.nan, np.nan], [92.11310000, 91.05430000, 90.95720000],
[35.4411, 34.1799, 34.0223],
[92.11310000, 91.05430000, 90.95720000]], dates, tickers),
'lookback_low':
pd.DataFrame([[np.nan, np.nan, np.nan], [34.1705, 92.453, 58.5107],
[15.67180000, 12.34530000, 34.05270000],
[27.18340000, 12.34530000, 23.29320000]], dates, tickers)
}
fn_correct_outputs = OrderedDict([
('long_short',
pd.DataFrame([[0, 0, 0], [-1, -1, -1], [1, 1, -1], [0, 0, 0]], dates,
tickers))
])
assert_output(fn, fn_inputs, fn_correct_outputs)
def test_get_signal_return(fn):
tickers = generate_random_tickers(3)
dates = generate_random_dates(5)
fn_inputs = {
'signal':
pd.DataFrame(
[[0, 0, 0], [-1, -1, -1], [1, 0, 0], [0, 0, 0], [0, 1, 0]], dates,
tickers),
'lookahead_returns':
pd.DataFrame(
[[0.88702896, 0.96521098, 0.65854789],
[1.13391240, 0.87420969, -0.53914925],
[0.35450805, -0.56900529, -0.64808965],
[0.38572896, -0.94655617, 0.123564379], [np.nan, np.nan, np.nan]],
dates, tickers)
}
fn_correct_outputs = OrderedDict([
('signal_return',
pd.DataFrame(
[[0, 0, 0], [-1.13391240, -0.87420969, 0.53914925],
[0.35450805, 0, 0], [0, 0, 0], [np.nan, np.nan, np.nan]], dates,
tickers))
])
assert_output(fn, fn_inputs, fn_correct_outputs)
def test_get_return_lookahead(fn):
tickers = generate_random_tickers(3)
dates = generate_random_dates(5)
fn_inputs = {
'close':
pd.DataFrame([[25.6788, 35.1392, 34.0527], [25.1884, 14.3453, 39.9373],
[62.3457, 92.2524, 65.7893], [78.2803, 34.3854, 23.2932],
[88.8725, 52.223, 34.4107]], dates, tickers),
'lookahead_prices':
pd.DataFrame([[62.34570000, 92.25240000, 65.78930000],
[78.28030000, 34.38540000, 23.29320000],
[88.87250000, 52.22300000, 34.41070000],
[np.nan, np.nan, np.nan], [np.nan, np.nan, np.nan]],
dates, tickers)
}
fn_correct_outputs = OrderedDict([
('lookahead_returns',
pd.DataFrame([[0.88702896, 0.96521098, 0.65854789],
[1.13391240, 0.87420969, -0.53914925],
[0.35450805, -0.56900529, -0.64808965],
[np.nan, np.nan, np.nan], [np.nan, np.nan, np.nan]],
dates, tickers))
])
assert_output(fn, fn_inputs, fn_correct_outputs)
def test_generate_positions(fn):
tickers = generate_random_tickers(5)
dates = generate_random_dates(6)
fn_inputs = {
'prices': pd.DataFrame(
[
[65.40757705426432, 27.556319958924323, 50.59935209411175, 56.274712269629134, 99.9873070881051],
[47.82126720752715, 56.812865745668375, 40.75685814634723, 27.469680989736023, 41.449858088448735],
[88.20038097315815, 45.573972499280494, 36.592711369868724, 21.36570423559795, 0.698919959739297],
[14.670236824202721, 49.557949251949054, 18.935364730808935, 23.163368660093298, 8.075599541367884],
[41.499140208637705, 9.75987296846733, 66.08677766062186, 37.927861417544385, 10.792730405945827],
[86.26923464863536, 32.12679487375028, 15.621592524570282, 77.1908860965619, 52.733950486350444]],
dates, tickers)}
fn_correct_outputs = OrderedDict([
(
'final_positions',
pd.DataFrame(
[
[30, 0, 30, 30, 30],
[0, 30, 0, 0, 0],
[30, 0, 0, 0, -10],
[-10, 0, -10, 0, -10],
[0, -10, 30, 0, -10],
[30, 0, -10, 30, 30]],
dates, tickers))])
assert_output(fn, fn_inputs, fn_correct_outputs)
def test_factor_betas(fn):
n_components = 3
dates = generate_random_dates(4)
assets = get_assets(3)
pca = PCA(n_components)
pca.fit(
pd.DataFrame([[0.21487253, 0.12342312, -0.13245215],
[0.23423439, -0.23434532, 1.67834324],
[0.23432445, -0.23563226, 0.23423523],
[0.24824535, -0.23523435, 0.36235236]], dates, assets))
fn_inputs = {
'pca': pca,
'factor_beta_indices': np.array(assets),
'factor_beta_columns': np.arange(n_components)
}
fn_correct_outputs = OrderedDict([
('factor_betas',
pd.DataFrame([[0.00590170, -0.07759542, 0.99696746],
[-0.13077609, 0.98836246, 0.07769983],
[0.99139436, 0.13083807, 0.00431461]],
fn_inputs['factor_beta_indices'],
fn_inputs['factor_beta_columns']))
])
assert_output(fn,
fn_inputs,
fn_correct_outputs,
check_parameter_changes=False)
def test_generate_dollar_volume_weights(fn):
tickers = generate_random_tickers(3)
dates = generate_random_dates(4)
fn_inputs = {
'close': pd.DataFrame(
[
[35.4411, 34.1799, 34.0223],
[92.1131, 91.0543, 90.9572],
[57.9708, 57.7814, 58.1982],
[34.1705, 92.453, 58.5107]],
dates, tickers),
'volume': pd.DataFrame(
[
[9.83683e+06, 1.78072e+07, 8.82982e+06],
[8.22427e+07, 6.85315e+07, 4.81601e+07],
[1.62348e+07, 1.30527e+07, 9.51201e+06],
[1.06742e+07, 5.68313e+07, 9.31601e+06]],
dates, tickers)}
fn_correct_outputs = OrderedDict([
(
'dollar_volume_weights',
pd.DataFrame(
[
[0.27719777, 0.48394253, 0.23885970],
[0.41632975, 0.34293308, 0.24073717],
[0.41848548, 0.33536102, 0.24615350],
[0.05917255, 0.85239760, 0.08842984]],
dates, tickers))])
assert_output(fn, fn_inputs, fn_correct_outputs)
def test_idiosyncratic_var_matrix(fn):
dates = generate_random_dates(4)
assets = get_assets(3)
fn_inputs = {
'returns': pd.DataFrame(
[
[ 0.02769242, 1.34872387, 0.23460972],
[-0.94728692, 0.68386883, -1.23987235],
[ 1.93769376, -0.48275934, 0.34957348],
[ 0.23985234, 0.35897345, 0.34598734]],
dates, assets),
'factor_returns': pd.DataFrame([
[-0.49503261, 1.45332369, -0.08980631],
[-1.87563271, 0.67894147, -1.11984992],
[-0.13027172, -0.49001128, 1.67259298],
[-0.25392567, 0.47320133, 0.04528734]],
dates),
'factor_betas': pd.DataFrame([
[ 0.00590170, -0.07759542, 0.99696746],
[-0.13077609, 0.98836246, 0.07769983],
[ 0.99139436, 0.13083807, 0.00431461]]),
'ann_factor': 250}
fn_correct_outputs = OrderedDict([
(
'idiosyncratic_var_matrix', pd.DataFrame(np.full([3,3], 0.0), assets, assets))])
assert_output(fn, fn_inputs, fn_correct_outputs)
def test_calculate_simple_moving_average(fn):
tickers = generate_random_tickers(5)
dates = generate_random_dates(6)
fn_inputs = {
'close': pd.DataFrame(
[
[21.050810483942833, 17.013843810658827, 10.984503755486879, 11.248093428369392, 12.961712733997235],
[15.63570258751384, 14.69054309070934, 11.353027688995159, 475.74195118202061, 11.959640427803022],
[482.34539247360806, 35.202580592515041, 3516.5416782257166, 66.405314327318209, 13.503960481087077],
[10.918933017418304, 17.9086438675435, 24.801265417692324, 12.488954191854916, 10.52435923388642],
[10.675971965144655, 12.749401436636365, 11.805257579935713, 21.539039489843024, 19.99766036804861],
[11.545495378369814, 23.981468434099405, 24.974763062186504, 36.031962102997689, 14.304332320024963]],
dates, tickers),
'rolling_window': 3}
fn_correct_outputs = OrderedDict([
(
'returns',
pd.DataFrame(
[
[np.nan, np.nan, np.nan, np.nan, np.nan],
[np.nan, np.nan, np.nan, np.nan, np.nan],
[173.01063518,22.30232250,1179.62640322,184.46511965,12.80843788],
[169.63334269,22.60058918,1184.23199044,184.87873990,11.99598671],
[167.98009915,21.95354197,1184.38273374,33.47776934,14.67532669],
[11.04680012,18.21317125,20.52709535,23.35331859,14.94211731]],
dates, tickers))])
assert_output(fn, fn_inputs, fn_correct_outputs)
def test_get_signal_return(fn):
tickers = generate_random_tickers(3)
dates = generate_random_dates(5)
fn_inputs = {
'signal': pd.DataFrame(
[
[0, 0, 0],
[-1, -1, -1],
[1, 0, 0],
[0, 0, 0],
[0, 1, 0]],
dates, tickers),
'lookahead_returns': pd.DataFrame(
[
[0.88702896, 0.96521098, 0.65854789],
[1.13391240, 0.87420969, -0.53914925],
[0.35450805, -0.56900529, -0.64808965],
[0.38572896, -0.94655617, 0.123564379],
[np.nan, np.nan, np.nan]],
dates, tickers)}
fn_correct_outputs = OrderedDict([
(
'signal_return',
pd.DataFrame(
[
[0, 0, 0],
[-1.13391240, -0.87420969, 0.53914925],
[0.35450805, 0, 0],
[0, 0, 0],
[np.nan, np.nan, np.nan]],
dates, tickers))])
assert_output(fn, fn_inputs, fn_correct_outputs)
def test_calculate_dividend_weights(fn):
tickers = generate_random_tickers(3)
dates = generate_random_dates(4)
fn_inputs = {
"dividends":
pd.DataFrame(
[[0.0, 0.0, 0.0], [0.0, 0.0, 0.1], [0.0, 1.0, 0.3],
[0.0, 0.2, 0.0]],
dates,
tickers,
)
}
fn_correct_outputs = OrderedDict([(
"dividend_weights",
pd.DataFrame(
[
[np.nan, np.nan, np.nan],
[0.00000000, 0.00000000, 1.00000000],
[0.00000000, 0.71428571, 0.28571429],
[0.00000000, 0.75000000, 0.25000000],
],
dates,
tickers,
),
)])
assert_output(fn, fn_inputs, fn_correct_outputs)
def test_get_lookahead_prices(fn):
tickers = generate_random_tickers(3)
dates = generate_random_dates(5)
fn_inputs = {
'close': pd.DataFrame(
[
[25.6788, 35.1392, 34.0527],
[25.1884, 14.3453, 39.9373],
[62.3457, 92.2524, 65.7893],
[78.2803, 34.3854, 23.2932],
[88.8725, 52.223, 34.4107]],
dates, tickers),
'lookahead_days': 2}
fn_correct_outputs = OrderedDict([
(
'lookahead_prices',
pd.DataFrame(
[
[62.34570000, 92.25240000, 65.78930000],
[78.28030000, 34.38540000, 23.29320000],
[88.87250000, 52.22300000, 34.41070000],
[np.nan, np.nan, np.nan],
[np.nan, np.nan, np.nan]],
dates, tickers))])
assert_output(fn, fn_inputs, fn_correct_outputs)
def test_factor_cov_matrix(fn):
dates = generate_random_dates(4)
fn_inputs = {
"factor_returns":
pd.DataFrame(
[
[-0.49503261, 1.45332369, -0.08980631],
[-1.87563271, 0.67894147, -1.11984992],
[-0.13027172, -0.49001128, 1.67259298],
[-0.25392567, 0.47320133, 0.04528734],
],
dates,
),
"ann_factor":
250,
}
fn_correct_outputs = OrderedDict([(
"factor_cov_matrix",
np.array([
[162.26559808, 0.0, 0.0],
[0.0, 159.86284454, 0.0],
[0.0, 0.0, 333.09785876],
]),
)])
assert_output(fn, fn_inputs, fn_correct_outputs)
def test_idiosyncratic_var_vector(fn):
dates = generate_random_dates(4)
assets = get_assets(3)
fn_inputs = {
"returns":
pd.DataFrame(
[
[0.02769242, 1.34872387, 0.23460972],
[-0.94728692, 0.68386883, -1.23987235],
[1.93769376, -0.48275934, 0.34957348],
[0.23985234, 0.35897345, 0.34598734],
],
dates,
assets,
),
"idiosyncratic_var_matrix":
pd.DataFrame(
[
[0.02272535, 0.0, 0.0],
[0.0, 0.05190083, 0.0],
[0.0, -0.49001128, 0.05431181],
],
assets,
assets,
),
}
fn_correct_outputs = OrderedDict([(
"idiosyncratic_var_vector",
pd.DataFrame([0.02272535, 0.05190083, 0.05431181], assets),
)])
assert_output(fn, fn_inputs, fn_correct_outputs)
def test_generate_weighted_returns(fn):
tickers = generate_random_tickers(3)
dates = generate_random_dates(4)
fn_inputs = {
'returns': pd.DataFrame(
[
[np.nan, np.nan, np.nan],
[1.59904743, 1.66397210, 1.67345829],
[-0.37065629, -0.36541822, -0.36015840],
[-0.41055669, 0.60004777, 0.00536958]],
dates, tickers),
'weights': pd.DataFrame(
[
[0.03777059, 0.04733924, 0.05197790],
[0.82074874, 0.48533938, 0.75792752],
[0.10196420, 0.05866016, 0.09578226],
[0.03951647, 0.40866122, 0.09431233]],
dates, tickers)}
fn_correct_outputs = OrderedDict([
(
'weighted_returns',
pd.DataFrame(
[
[np.nan, np.nan, np.nan],
[1.31241616, 0.80759119, 1.26836009],
[-0.03779367, -0.02143549, -0.03449679],
[-0.01622375, 0.24521625, 0.00050642]],
dates, tickers))])
assert_output(fn, fn_inputs, fn_correct_outputs)
def test_factor_betas(fn):
n_components = 3
dates = generate_random_dates(4)
assets = get_assets(3)
pca = PCA(n_components)
pca.fit(pd.DataFrame(
[
[0.21487253, 0.12342312, -0.13245215],
[0.23423439, -0.23434532, 1.67834324],
[0.23432445, -0.23563226, 0.23423523],
[0.24824535, -0.23523435, 0.36235236]],
dates, assets))
fn_inputs = {
'pca': pca,
'factor_beta_indices': np.array(assets),
'factor_beta_columns': np.arange(n_components)}
fn_correct_outputs = OrderedDict([
(
'factor_betas', pd.DataFrame(
[
[ 0.00590170, -0.07759542, 0.99696746],
[-0.13077609, 0.98836246, 0.07769983],
[ 0.99139436, 0.13083807, 0.00431461]],
fn_inputs['factor_beta_indices'],
fn_inputs['factor_beta_columns']))])
assert_output(fn, fn_inputs, fn_correct_outputs, check_parameter_changes=False)
def test_get_high_lows_lookback(fn):
tickers = generate_random_tickers(3)
dates = generate_random_dates(4)
fn_inputs = {
"high":
pd.DataFrame(
[
[35.4411, 34.1799, 34.0223],
[92.1131, 91.0543, 90.9572],
[57.9708, 57.7814, 58.1982],
[34.1705, 92.453, 58.5107],
],
dates,
tickers,
),
"low":
pd.DataFrame(
[
[15.6718, 75.1392, 34.0527],
[27.1834, 12.3453, 95.9373],
[28.2503, 24.2854, 23.2932],
[86.3725, 32.223, 38.4107],
],
dates,
tickers,
),
"lookback_days":
2,
}
fn_correct_outputs = OrderedDict([
(
"lookback_high",
pd.DataFrame(
[
[np.nan, np.nan, np.nan],
[np.nan, np.nan, np.nan],
[92.11310000, 91.05430000, 90.95720000],
[92.11310000, 91.05430000, 90.95720000],
],
dates,
tickers,
),
),
(
"lookback_low",
pd.DataFrame(
[
[np.nan, np.nan, np.nan],
[np.nan, np.nan, np.nan],
[15.67180000, 12.34530000, 34.05270000],
[27.18340000, 12.34530000, 23.29320000],
],
dates,
tickers,
),
),
])
assert_output(fn, fn_inputs, fn_correct_outputs)
def test_idiosyncratic_var_matrix(fn):
dates = generate_random_dates(4)
assets = get_assets(3)
fn_inputs = {
'returns':
pd.DataFrame([[0.02769242, 1.34872387, 0.23460972],
[-0.94728692, 0.68386883, -1.23987235],
[1.93769376, -0.48275934, 0.34957348],
[0.23985234, 0.35897345, 0.34598734]], dates, assets),
'factor_returns':
pd.DataFrame([[-0.49503261, 1.45332369, -0.08980631],
[-1.87563271, 0.67894147, -1.11984992],
[-0.13027172, -0.49001128, 1.67259298],
[-0.25392567, 0.47320133, 0.04528734]], dates),
'factor_betas':
pd.DataFrame([[0.00590170, -0.07759542, 0.99696746],
[-0.13077609, 0.98836246, 0.07769983],
[0.99139436, 0.13083807, 0.00431461]]),
'ann_factor':
250
}
fn_correct_outputs = OrderedDict([('idiosyncratic_var_matrix',
pd.DataFrame(np.full([3, 3], 0.0),
assets, assets))])
assert_output(fn, fn_inputs, fn_correct_outputs)
def test_date_top_industries(fn):
tickers = generate_random_tickers(10)
dates = generate_random_dates(2)
fn_inputs = {
'prices':
pd.DataFrame(
[[
21.050810483942833, 17.013843810658827, 10.984503755486879,
11.248093428369392, 12.961712733997235, 482.34539247360806,
35.202580592515041, 3516.5416782257166, 66.405314327318209,
13.503960481087077
],
[
15.63570258751384, 14.69054309070934, 11.353027688995159,
475.74195118202061, 11.959640427803022, 10.918933017418304,
17.9086438675435, 24.801265417692324, 12.488954191854916,
15.63570258751384
]], dates, tickers),
'sector':
pd.Series([
'ENERGY', 'MATERIALS', 'ENERGY', 'ENERGY', 'TELECOM', 'FINANCIALS',
'TECHNOLOGY', 'HEALTH', 'MATERIALS', 'REAL ESTATE'
], tickers),
'date':
dates[-1],
'top_n':
4
}
fn_correct_outputs = OrderedDict([('top_industries',
{'ENERGY', 'HEALTH', 'TECHNOLOGY'})])
assert_output(fn, fn_inputs, fn_correct_outputs)
def test_generate_returns(fn):
tickers = generate_random_tickers(3)
dates = generate_random_dates(4)
fn_inputs = {
"prices":
pd.DataFrame(
[
[35.4411, 34.1799, 34.0223],
[92.1131, 91.0543, 90.9572],
[57.9708, 57.7814, 58.1982],
[34.1705, 92.453, 58.5107],
],
dates,
tickers,
)
}
fn_correct_outputs = OrderedDict([(
"returns",
pd.DataFrame(
[
[np.nan, np.nan, np.nan],
[1.59904743, 1.66397210, 1.67345829],
[-0.37065629, -0.36541822, -0.36015840],
[-0.41055669, 0.60004777, 0.00536958],
],
dates,
tickers,
),
)])
assert_output(fn, fn_inputs, fn_correct_outputs)
def test_get_covariance_returns(fn):
tickers = generate_random_tickers(3)
dates = generate_random_dates(4)
fn_inputs = {
"returns":
pd.DataFrame(
[
[np.nan, np.nan, np.nan],
[1.59904743, 1.66397210, 1.67345829],
[-0.37065629, -0.36541822, -0.36015840],
[-0.41055669, 0.60004777, 0.00536958],
],
dates,
tickers,
)
}
fn_correct_outputs = OrderedDict([(
"returns_covariance",
np.array([
[0.89856076, 0.7205586, 0.8458721],
[0.7205586, 0.78707297, 0.76450378],
[0.8458721, 0.76450378, 0.83182775],
]),
)])
assert_output(fn, fn_inputs, fn_correct_outputs)
def test_get_return_lookahead(fn):
tickers = generate_random_tickers(3)
dates = generate_random_dates(5)
fn_inputs = {
'close': pd.DataFrame(
[
[25.6788, 35.1392, 34.0527],
[25.1884, 14.3453, 39.9373],
[62.3457, 92.2524, 65.7893],
[78.2803, 34.3854, 23.2932],
[88.8725, 52.223, 34.4107]],
dates, tickers),
'lookahead_prices': pd.DataFrame(
[
[62.34570000, 92.25240000, 65.78930000],
[78.28030000, 34.38540000, 23.29320000],
[88.87250000, 52.22300000, 34.41070000],
[np.nan, np.nan, np.nan],
[np.nan, np.nan, np.nan]],
dates, tickers)}
fn_correct_outputs = OrderedDict([
(
'lookahead_returns',
pd.DataFrame(
[
[0.88702896, 0.96521098, 0.65854789],
[1.13391240, 0.87420969, -0.53914925],
[0.35450805, -0.56900529, -0.64808965],
[np.nan, np.nan, np.nan],
[np.nan, np.nan, np.nan]],
dates, tickers))])
assert_output(fn, fn_inputs, fn_correct_outputs)
def test_non_overlapping_estimators(fn):
n_estimators = 3
columns = ['test column 1', 'test column 2']
dates = generate_random_dates(8)
assets = get_assets(3)
index = pd.MultiIndex.from_product([dates, assets])
noise = np.random.RandomState(0).random_sample([len(index)]) * len(index)
values = np.arange(len(index) * len(columns)).reshape([len(columns), len(index)]).T
targets = np.sum(values, axis=-1) + noise
classifiers = [
RandomForestRegressor(300, oob_score=True, n_jobs=-1, random_state=101)
for _ in range(n_estimators)]
fn_inputs = {
'x': pd.DataFrame(values, index, columns),
'y': pd.Series(targets, index),
'classifiers': classifiers,
'n_skip_samples': 3}
random_forest_regressor_fit = RandomForestRegressor.fit
with patch.object(RandomForestRegressor, 'fit', autospec=True) as mock_fit:
mock_fit.side_effect = random_forest_regressor_fit
fn_return_value = fn(**fn_inputs)
assert_structure(fn_return_value, [RandomForestRegressor for _ in range(n_estimators)], 'PCA')
for classifier in fn_return_value:
try:
classifier.fit.assert_called()
except AssertionError:
raise Exception('Test Failure: RandomForestRegressor.fit not called on all classifiers')
def test_fit_pca(fn):
dates = generate_random_dates(4)
assets = get_assets(3)
fn_inputs = {
'returns':
pd.DataFrame([[0.02769242, 1.34872387, 0.23460972],
[-0.94728692, 0.68386883, -1.23987235],
[1.93769376, -0.48275934, 0.34957348],
[0.23985234, 0.35897345, 0.34598734]], dates, assets),
'num_factor_exposures':
2,
'svd_solver':
'full'
}
fn_correct_values = {
'PCA':
PCA(),
'PCA.components_':
np.array([[0.81925896, -0.40427891, 0.40666118],
[-0.02011128, 0.68848693, 0.72496985]])
}
pca_fit = PCA.fit
with patch.object(PCA, 'fit', autospec=True) as mock_fit:
mock_fit.side_effect = pca_fit
fn_return_value = fn(**fn_inputs)
assert_structure(fn_return_value, fn_correct_values['PCA'], 'PCA')
try:
# print(dir(fn_return_value.fit))
fn_return_value.fit.assert_called()
# old python:
# fn_return_value.fit.assert_any_call()
except AssertionError:
raise Exception('Test Failure: PCA.fit not called')
try:
fn_return_value.fit.assert_called_with(self=fn_return_value,
X=fn_inputs['returns'])
except Exception:
raise Exception(
'Test Failure: PCA.fit called with the wrong arguments')
assert_structure(fn_return_value.components_,
fn_correct_values['PCA.components_'],
'PCA.components_')
if not does_data_match(fn_return_value.components_,
fn_correct_values['PCA.components_']):
raise Exception('Test Failure: PCA not fitted correctly\n\n'
'PCA.components_:\n'
'{}\n\n'
'Expected PCA.components_:\n'
'{}'.format(fn_return_value.components_,
fn_correct_values['PCA.components_']))
def test_factor_returns(fn):
n_components = 3
dates = generate_random_dates(4)
assets = get_assets(3)
pca = PCA(n_components)
pca.fit(
pd.DataFrame(
[
[0.21487253, 0.12342312, -0.13245215],
[0.23423439, -0.23434532, 1.67834324],
[0.23432445, -0.23563226, 0.23423523],
[0.24824535, -0.23523435, 0.36235236],
],
dates,
assets,
))
fn_inputs = {
"pca":
pca,
"returns":
pd.DataFrame(
[
[0.02769242, 1.34872387, 0.23460972],
[-0.94728692, 0.68386883, -1.23987235],
[1.93769376, -0.48275934, 0.34957348],
[0.23985234, 0.35897345, 0.34598734],
],
dates,
assets,
),
"factor_return_indices":
np.array(dates),
"factor_return_columns":
np.arange(n_components),
}
fn_correct_outputs = OrderedDict([(
"factor_returns",
pd.DataFrame(
[
[-0.49503261, 1.45332369, -0.08980631],
[-1.87563271, 0.67894147, -1.11984992],
[-0.13027172, -0.49001128, 1.67259298],
[-0.25392567, 0.47320133, 0.04528734],
],
fn_inputs["factor_return_indices"],
fn_inputs["factor_return_columns"],
),
)])
assert_output(fn,
fn_inputs,
fn_correct_outputs,
check_parameter_changes=False)
def test_tracking_error(fn):
dates = generate_random_dates(4)
fn_inputs = {
'benchmark_returns_by_date':
pd.Series([np.nan, 0.99880148, 0.99876653, 1.00024411], dates),
'etf_returns_by_date':
pd.Series([np.nan, 0.63859274, 0.93475823, 2.57295727], dates)
}
fn_correct_outputs = OrderedDict([('tracking_error', 16.5262431971)])
assert_output(fn, fn_inputs, fn_correct_outputs)
def test_rebalance_portfolio(fn):
tickers = generate_random_tickers(3)
dates = generate_random_dates(11)
fn_inputs = {
'returns':
pd.DataFrame(
[[np.nan, np.nan, np.nan], [-0.02202381, 0.02265285, 0.01441961],
[0.01947657, 0.00551985, 0.00047382],
[0.00537313, -0.00803232, 0.01160313],
[0.00593824, -0.00567773, 0.02247191],
[0.02479339, 0.01758824, -0.00824176],
[-0.0109447, -0.00383568, 0.01361958],
[0.01164822, 0.01558719, 0.00614894],
[0.0109384, -0.00182079, 0.02900868],
[0.01138952, 0.00218049, -0.00954495],
[0.0106982, 0.00644535, -0.01815329]], dates, tickers),
'index_weights':
pd.DataFrame([[0.00449404, 0.11586048, 0.00359727],
[
0.00403487,
0.12534048,
0.0034428,
], [0.00423485, 0.12854258, 0.00347404],
[0.00395679, 0.1243466, 0.00335064],
[0.00368729, 0.11750295, 0.00333929],
[0.00369562, 0.11447422, 0.00325973],
[
0.00379612,
0.11088075,
0.0031734,
], [0.00366501, 0.10806014, 0.00314648],
[0.00361268, 0.10376514, 0.00323257],
[0.00358844, 0.10097531, 0.00319009],
[0.00362045, 0.09791232, 0.00318071]], dates, tickers),
'shift_size':
2,
'chunk_size':
4
}
fn_correct_outputs = OrderedDict([('all_rebalance_weights', [
np.array([0.29341237, 0.41378419, 0.29280344]),
np.array([0.29654088, 0.40731481, 0.29614432]),
np.array([0.29868214, 0.40308791, 0.29822995]),
np.array([0.30100044, 0.39839644, 0.30060312])
])])
assert_output(fn, fn_inputs, fn_correct_outputs)
def test_tracking_error(fn):
dates = generate_random_dates(4)
fn_inputs = {
'benchmark_returns_by_date': pd.Series(
[np.nan, 0.99880148, 0.99876653, 1.00024411],
dates),
'etf_returns_by_date': pd.Series(
[np.nan, 0.63859274, 0.93475823, 2.57295727],
dates)}
fn_correct_outputs = OrderedDict([
(
'tracking_error',
16.5262431971)])
assert_output(fn, fn_inputs, fn_correct_outputs)
def test_filter_signals(fn):
tickers = generate_random_tickers(3)
dates = generate_random_dates(10)
fn_inputs = {
"signal":
pd.DataFrame(
[
[0, 0, 0],
[-1, -1, -1],
[1, 0, -1],
[0, 0, 0],
[1, 0, 0],
[0, 1, 0],
[0, 0, 1],
[0, -1, 1],
[-1, 0, 0],
[0, 0, 0],
],
dates,
tickers,
),
"lookahead_days":
3,
}
fn_correct_outputs = OrderedDict([(
"filtered_signal",
pd.DataFrame(
[
[0, 0, 0],
[-1, -1, -1],
[1, 0, 0],
[0, 0, 0],
[0, 0, 0],
[0, 1, 0],
[0, 0, 1],
[0, -1, 0],
[-1, 0, 0],
[0, 0, 0],
],
dates,
tickers,
),
)])
assert_output(fn, fn_inputs, fn_correct_outputs)
def test_fit_pca(fn):
dates = generate_random_dates(4)
assets = get_assets(3)
fn_inputs = {
'returns': pd.DataFrame(
[
[0.02769242, 1.34872387, 0.23460972],
[-0.94728692, 0.68386883, -1.23987235],
[1.93769376, -0.48275934, 0.34957348],
[0.23985234, 0.35897345, 0.34598734]],
dates, assets),
'num_factor_exposures': 2,
'svd_solver': 'full'}
fn_correct_values = {
'PCA': PCA(),
'PCA.components_': np.array([
[0.81925896, -0.40427891, 0.40666118],
[-0.02011128, 0.68848693, 0.72496985]])}
pca_fit = PCA.fit
with patch.object(PCA, 'fit', autospec=True) as mock_fit:
mock_fit.side_effect = pca_fit
fn_return_value = fn(**fn_inputs)
assert_structure(fn_return_value, fn_correct_values['PCA'], 'PCA')
try:
fn_return_value.fit.assert_called()
except AssertionError:
raise Exception('Test Failure: PCA.fit not called')
try:
fn_return_value.fit.assert_called_with(self=fn_return_value, X=fn_inputs['returns'])
except Exception:
raise Exception('Test Failure: PCA.fit called with the wrong arguments')
assert_structure(fn_return_value.components_, fn_correct_values['PCA.components_'], 'PCA.components_')
if not does_data_match(fn_return_value.components_, fn_correct_values['PCA.components_']):
raise Exception('Test Failure: PCA not fitted correctly\n\n'
'PCA.components_:\n'
'{}\n\n'
'Expected PCA.components_:\n'
'{}'.format(fn_return_value.components_, fn_correct_values['PCA.components_']))
def test_rebalance_portfolio(fn):
tickers = generate_random_tickers(3)
dates = generate_random_dates(11)
fn_inputs = {
'returns': pd.DataFrame(
[
[np.nan, np.nan, np.nan],
[-0.02202381, 0.02265285, 0.01441961],
[0.01947657, 0.00551985, 0.00047382],
[0.00537313, -0.00803232, 0.01160313],
[0.00593824, -0.00567773, 0.02247191],
[0.02479339, 0.01758824, -0.00824176],
[-0.0109447, -0.00383568, 0.01361958],
[0.01164822, 0.01558719, 0.00614894],
[0.0109384, -0.00182079, 0.02900868],
[0.01138952, 0.00218049, -0.00954495],
[0.0106982, 0.00644535, -0.01815329]],
dates, tickers),
'index_weights': pd.DataFrame(
[
[0.00449404, 0.11586048, 0.00359727],
[0.00403487, 0.12534048, 0.0034428, ],
[0.00423485, 0.12854258, 0.00347404],
[0.00395679, 0.1243466, 0.00335064],
[0.00368729, 0.11750295, 0.00333929],
[0.00369562, 0.11447422, 0.00325973],
[0.00379612, 0.11088075, 0.0031734, ],
[0.00366501, 0.10806014, 0.00314648],
[0.00361268, 0.10376514, 0.00323257],
[0.00358844, 0.10097531, 0.00319009],
[0.00362045, 0.09791232, 0.00318071]],
dates, tickers),
'shift_size': 2,
'chunk_size': 4}
fn_correct_outputs = OrderedDict([
(
'all_rebalance_weights',
[
np.array([0.29341237, 0.41378419, 0.29280344]),
np.array([0.29654088, 0.40731481, 0.29614432]),
np.array([0.29868214, 0.40308791, 0.29822995]),
np.array([0.30100044, 0.39839644, 0.30060312])]
)])
assert_output(fn, fn_inputs, fn_correct_outputs)
def test_calculate_cumulative_returns(fn):
tickers = generate_random_tickers(3)
dates = generate_random_dates(4)
fn_inputs = {
'returns':
pd.DataFrame(
[[np.nan, np.nan, np.nan], [1.59904743, 1.66397210, 1.67345829],
[-0.37065629, -0.36541822, -0.36015840],
[-0.41055669, 0.60004777, 0.00536958]], dates, tickers)
}
fn_correct_outputs = OrderedDict([
('cumulative_returns',
pd.Series([np.nan, 5.93647782, -0.57128454, -0.68260542], dates))
])
assert_output(fn, fn_inputs, fn_correct_outputs)
def test_generate_weighted_returns(fn):
tickers = generate_random_tickers(3)
dates = generate_random_dates(4)
fn_inputs = {
"returns": pd.DataFrame(
[
[np.nan, np.nan, np.nan],
[1.59904743, 1.66397210, 1.67345829],
[-0.37065629, -0.36541822, -0.36015840],
[-0.41055669, 0.60004777, 0.00536958],
],
dates,
tickers,
),
"weights": pd.DataFrame(
[
[0.03777059, 0.04733924, 0.05197790],
[0.82074874, 0.48533938, 0.75792752],
[0.10196420, 0.05866016, 0.09578226],
[0.03951647, 0.40866122, 0.09431233],
],
dates,
tickers,
),
}
fn_correct_outputs = OrderedDict(
[
(
"weighted_returns",
pd.DataFrame(
[
[np.nan, np.nan, np.nan],
[1.31241616, 0.80759119, 1.26836009],
[-0.03779367, -0.02143549, -0.03449679],
[-0.01622375, 0.24521625, 0.00050642],
],
dates,
tickers,
),
)
]
)
assert_output(fn, fn_inputs, fn_correct_outputs)
def test_generate_dollar_volume_weights(fn):
tickers = generate_random_tickers(3)
dates = generate_random_dates(4)
fn_inputs = {
"close": pd.DataFrame(
[
[35.4411, 34.1799, 34.0223],
[92.1131, 91.0543, 90.9572],
[57.9708, 57.7814, 58.1982],
[34.1705, 92.453, 58.5107],
],
dates,
tickers,
),
"volume": pd.DataFrame(
[
[9.83683e06, 1.78072e07, 8.82982e06],
[8.22427e07, 6.85315e07, 4.81601e07],
[1.62348e07, 1.30527e07, 9.51201e06],
[1.06742e07, 5.68313e07, 9.31601e06],
],
dates,
tickers,
),
}
fn_correct_outputs = OrderedDict(
[
(
"dollar_volume_weights",
pd.DataFrame(
[
[0.27719777, 0.48394253, 0.23885970],
[0.41632975, 0.34293308, 0.24073717],
[0.41848548, 0.33536102, 0.24615350],
[0.05917255, 0.85239760, 0.08842984],
],
dates,
tickers,
),
)
]
)
assert_output(fn, fn_inputs, fn_correct_outputs)
def test_calculate_returns(fn):
tickers = generate_random_tickers(5)
dates = generate_random_dates(6)
fn_inputs = {
'close':
pd.DataFrame(
[[
21.050810483942833, 17.013843810658827, 10.984503755486879,
11.248093428369392, 12.961712733997235
],
[
15.63570258751384, 14.69054309070934, 11.353027688995159,
475.74195118202061, 11.959640427803022
],
[
482.34539247360806, 35.202580592515041, 3516.5416782257166,
66.405314327318209, 13.503960481087077
],
[
10.918933017418304, 17.9086438675435, 24.801265417692324,
12.488954191854916, 10.52435923388642
],
[
10.675971965144655, 12.749401436636365, 11.805257579935713,
21.539039489843024, 19.99766036804861
],
[
11.545495378369814, 23.981468434099405, 24.974763062186504,
36.031962102997689, 14.304332320024963
]], dates, tickers)
}
fn_correct_outputs = OrderedDict([
('returns',
pd.DataFrame([
[np.nan, np.nan, np.nan, np.nan, np.nan],
[-0.25723988, -0.13655355, 0.03354944, 41.29534136, -0.07731018],
[29.84897463, 1.39627496, 308.74483411, -0.86041737, 0.12912763],
[-0.97736283, -0.49126900, -0.99294726, -0.81192839, -0.22064647],
[-0.02225135, -0.28808672, -0.52400584, 0.72464717, 0.90013092],
[0.08144677, 0.88098779, 1.11556274, 0.67286764, -0.28469971]
], dates, tickers))
])
assert_output(fn, fn_inputs, fn_correct_outputs)
def test_sharpe_ratio(fn):
dates = generate_random_dates(4)
factor_names = ['Factor {}'.format(i) for i in range(3)]
fn_inputs = {
'factor_returns': pd.DataFrame(
[
[ 0.00069242, 0.00072387, 0.00002972],
[-0.00028692, 0.00086883, -0.00007235],
[-0.00066376, -0.00045934, 0.00007348],
[ 0.00085234, 0.00093345, 0.00008734]],
dates, factor_names),
'annualization_factor': 16.0}
fn_correct_outputs = OrderedDict([
(
'sharpe_ratio', pd.Series([3.21339895, 12.59157330, 6.54485802], factor_names))])
assert_output(fn, fn_inputs, fn_correct_outputs)
def test_factor_cov_matrix(fn):
dates = generate_random_dates(4)
fn_inputs = {
'factor_returns': pd.DataFrame([
[-0.49503261, 1.45332369, -0.08980631],
[-1.87563271, 0.67894147, -1.11984992],
[-0.13027172, -0.49001128, 1.67259298],
[-0.25392567, 0.47320133, 0.04528734]],
dates),
'ann_factor': 250}
fn_correct_outputs = OrderedDict([
(
'factor_cov_matrix', np.array([
[162.26559808, 0.0, 0.0],
[0.0, 159.86284454, 0.0],
[0.0, 0.0, 333.09785876]]))])
assert_output(fn, fn_inputs, fn_correct_outputs)
def test_get_high_lows_lookback(fn):
tickers = generate_random_tickers(3)
dates = generate_random_dates(4)
fn_inputs = {
'high': pd.DataFrame(
[
[35.4411, 34.1799, 34.0223],
[92.1131, 91.0543, 90.9572],
[57.9708, 57.7814, 58.1982],
[34.1705, 92.453, 58.5107]],
dates, tickers),
'low': pd.DataFrame(
[
[15.6718, 75.1392, 34.0527],
[27.1834, 12.3453, 95.9373],
[28.2503, 24.2854, 23.2932],
[86.3725, 32.223, 38.4107]],
dates, tickers),
'lookback_days': 2}
fn_correct_outputs = OrderedDict([
(
'lookback_high',
pd.DataFrame(
[
[np.nan, np.nan, np.nan],
[np.nan, np.nan, np.nan],
[92.11310000, 91.05430000, 90.95720000],
[92.11310000, 91.05430000, 90.95720000]],
dates, tickers)),
(
'lookback_low',
pd.DataFrame(
[
[np.nan, np.nan, np.nan],
[np.nan, np.nan, np.nan],
[15.67180000, 12.34530000, 34.05270000],
[27.18340000, 12.34530000, 23.29320000]],
dates, tickers))
])
assert_output(fn, fn_inputs, fn_correct_outputs)
def test_calculate_cumulative_returns(fn):
tickers = generate_random_tickers(3)
dates = generate_random_dates(4)
fn_inputs = {
'returns': pd.DataFrame(
[
[np.nan, np.nan, np.nan],
[1.59904743, 1.66397210, 1.67345829],
[-0.37065629, -0.36541822, -0.36015840],
[-0.41055669, 0.60004777, 0.00536958]],
dates, tickers)}
fn_correct_outputs = OrderedDict([
(
'cumulative_returns',
pd.Series(
[np.nan, 5.93647782, -0.57128454, -0.68260542],
dates))])
assert_output(fn, fn_inputs, fn_correct_outputs)
def test_get_covariance_returns(fn):
tickers = generate_random_tickers(3)
dates = generate_random_dates(4)
fn_inputs = {
'returns': pd.DataFrame(
[
[np.nan, np.nan, np.nan],
[1.59904743, 1.66397210, 1.67345829],
[-0.37065629, -0.36541822, -0.36015840],
[-0.41055669, 0.60004777, 0.00536958]],
dates, tickers)}
fn_correct_outputs = OrderedDict([(
'returns_covariance',
np.array(
[
[0.89856076, 0.7205586, 0.8458721],
[0.7205586, 0.78707297, 0.76450378],
[0.8458721, 0.76450378, 0.83182775]]))])
assert_output(fn, fn_inputs, fn_correct_outputs)
def test_calculate_arithmetic_rate_of_return(fn):
tickers = generate_random_tickers(5)
dates = generate_random_dates(6)
fn_inputs = {
'close':
pd.DataFrame(
[[
21.050810483942833, 17.013843810658827, 10.984503755486879,
11.248093428369392, 12.961712733997235
],
[
15.63570258751384, 14.69054309070934, 11.353027688995159,
475.74195118202061, 11.959640427803022
],
[
482.34539247360806, 35.202580592515041, 3516.5416782257166,
66.405314327318209, 13.503960481087077
],
[
10.918933017418304, 17.9086438675435, 24.801265417692324,
12.488954191854916, 10.52435923388642
],
[
10.675971965144655, 12.749401436636365, 11.805257579935713,
21.539039489843024, 19.99766036804861
],
[
11.545495378369814, 23.981468434099405, 24.974763062186504,
36.031962102997689, 14.304332320024963
]], dates, tickers)
}
fn_correct_outputs = OrderedDict([
('arithmetic_returns',
pd.Series(
[4.77892789, 0.22689225, 51.39616553, 6.83675173, 0.07443370],
tickers))
])
assert_output(fn, fn_inputs, fn_correct_outputs)
def test_filter_signals(fn):
tickers = generate_random_tickers(3)
dates = generate_random_dates(10)
fn_inputs = {
'signal':
pd.DataFrame(
[[0, 0, 0], [-1, -1, -1], [1, 0, -1], [0, 0, 0], [1, 0, 0],
[0, 1, 0], [0, 0, 1], [0, -1, 1], [-1, 0, 0], [0, 0, 0]], dates,
tickers),
'lookahead_days':
3
}
fn_correct_outputs = OrderedDict([
('filtered_signal',
pd.DataFrame(
[[0, 0, 0], [-1, -1, -1], [1, 0, 0], [0, 0, 0], [0, 0, 0],
[0, 1, 0], [0, 0, 1], [0, -1, 0], [-1, 0, 0], [0, 0, 0]], dates,
tickers))
])
assert_output(fn, fn_inputs, fn_correct_outputs)
def test_get_lookahead_prices(fn):
tickers = generate_random_tickers(3)
dates = generate_random_dates(5)
fn_inputs = {
'close':
pd.DataFrame([[25.6788, 35.1392, 34.0527], [25.1884, 14.3453, 39.9373],
[62.3457, 92.2524, 65.7893], [78.2803, 34.3854, 23.2932],
[88.8725, 52.223, 34.4107]], dates, tickers),
'lookahead_days':
2
}
fn_correct_outputs = OrderedDict([
('lookahead_prices',
pd.DataFrame([[62.34570000, 92.25240000, 65.78930000],
[78.28030000, 34.38540000, 23.29320000],
[88.87250000, 52.22300000, 34.41070000],
[np.nan, np.nan, np.nan], [np.nan, np.nan, np.nan]],
dates, tickers))
])
assert_output(fn, fn_inputs, fn_correct_outputs)
def test_idiosyncratic_var_vector(fn):
dates = generate_random_dates(4)
assets = get_assets(3)
fn_inputs = {
'returns': pd.DataFrame(
[
[ 0.02769242, 1.34872387, 0.23460972],
[-0.94728692, 0.68386883, -1.23987235],
[ 1.93769376, -0.48275934, 0.34957348],
[ 0.23985234, 0.35897345, 0.34598734]],
dates, assets),
'idiosyncratic_var_matrix': pd.DataFrame([
[0.02272535, 0.0, 0.0],
[0.0, 0.05190083, 0.0],
[0.0, -0.49001128, 0.05431181]],
assets, assets),}
fn_correct_outputs = OrderedDict([
(
'idiosyncratic_var_vector', pd.DataFrame([0.02272535, 0.05190083, 0.05431181], assets))])
assert_output(fn, fn_inputs, fn_correct_outputs)
def test_get_long_short(fn):
tickers = generate_random_tickers(3)
dates = generate_random_dates(4)
fn_inputs = {
'close': pd.DataFrame(
[
[25.6788, 35.1392, 34.0527],
[25.1884, 14.3453, 39.9373],
[78.2803, 34.3854, 23.2932],
[88.8725, 52.223, 34.4107]],
dates, tickers),
'lookback_high': pd.DataFrame(
[
[np.nan, np.nan, np.nan],
[92.11310000, 91.05430000, 90.95720000],
[35.4411, 34.1799, 34.0223],
[92.11310000, 91.05430000, 90.95720000]],
dates, tickers),
'lookback_low': pd.DataFrame(
[
[np.nan, np.nan, np.nan],
[34.1705, 92.453, 58.5107],
[15.67180000, 12.34530000, 34.05270000],
[27.18340000, 12.34530000, 23.29320000]],
dates, tickers)}
fn_correct_outputs = OrderedDict([
(
'long_short',
pd.DataFrame(
[
[0, 0, 0],
[-1, -1, -1],
[1, 1, -1],
[0, 0, 0]],
dates, tickers))])
assert_output(fn, fn_inputs, fn_correct_outputs)
def test_factor_returns(fn):
n_components = 3
dates = generate_random_dates(4)
assets = get_assets(3)
pca = PCA(n_components)
pca.fit(pd.DataFrame(
[
[0.21487253, 0.12342312, -0.13245215],
[0.23423439, -0.23434532, 1.67834324],
[0.23432445, -0.23563226, 0.23423523],
[0.24824535, -0.23523435, 0.36235236]],
dates, assets))
fn_inputs = {
'pca': pca,
'returns': pd.DataFrame(
[
[0.02769242, 1.34872387, 0.23460972],
[-0.94728692, 0.68386883, -1.23987235],
[1.93769376, -0.48275934, 0.34957348],
[0.23985234, 0.35897345, 0.34598734]],
dates, assets),
'factor_return_indices': np.array(dates),
'factor_return_columns': np.arange(n_components)}
fn_correct_outputs = OrderedDict([
(
'factor_returns', pd.DataFrame(
[
[-0.49503261, 1.45332369, -0.08980631],
[-1.87563271, 0.67894147, -1.11984992],
[-0.13027172, -0.49001128, 1.67259298],
[-0.25392567, 0.47320133, 0.04528734]],
fn_inputs['factor_return_indices'],
fn_inputs['factor_return_columns']))])
assert_output(fn, fn_inputs, fn_correct_outputs, check_parameter_changes=False)
def test_filter_signals(fn):
tickers = generate_random_tickers(3)
dates = generate_random_dates(10)
fn_inputs = {
'signal': pd.DataFrame(
[
[0, 0, 0],
[-1, -1, -1],
[1, 0, -1],
[0, 0, 0],
[1, 0, 0],
[0, 1, 0],
[0, 0, 1],
[0, -1, 1],
[-1, 0, 0],
[0, 0, 0]],
dates, tickers),
'lookahead_days': 3}
fn_correct_outputs = OrderedDict([
(
'filtered_signal',
pd.DataFrame(
[
[0, 0, 0],
[-1, -1, -1],
[1, 0, 0],
[0, 0, 0],
[0, 0, 0],
[0, 1, 0],
[0, 0, 1],
[0, -1, 0],
[-1, 0, 0],
[0, 0, 0]],
dates, tickers))])
assert_output(fn, fn_inputs, fn_correct_outputs)
def test_calculate_dividend_weights(fn):
tickers = generate_random_tickers(3)
dates = generate_random_dates(4)
fn_inputs = {
'dividends': pd.DataFrame(
[
[0.0, 0.0, 0.0],
[0.0, 0.0, 0.1],
[0.0, 1.0, 0.3],
[0.0, 0.2, 0.0]],
dates, tickers)}
fn_correct_outputs = OrderedDict([
(
'dividend_weights',
pd.DataFrame(
[
[np.nan, np.nan, np.nan],
[0.00000000, 0.00000000, 1.00000000],
[0.00000000, 0.71428571, 0.28571429],
[0.00000000, 0.75000000, 0.25000000]],
dates, tickers))])
assert_output(fn, fn_inputs, fn_correct_outputs)
def test_generate_returns(fn):
tickers = generate_random_tickers(3)
dates = generate_random_dates(4)
fn_inputs = {
'prices': pd.DataFrame(
[
[35.4411, 34.1799, 34.0223],
[92.1131, 91.0543, 90.9572],
[57.9708, 57.7814, 58.1982],
[34.1705, 92.453, 58.5107]],
dates, tickers)}
fn_correct_outputs = OrderedDict([
(
'returns',
pd.DataFrame(
[
[np.nan, np.nan, np.nan],
[1.59904743, 1.66397210, 1.67345829],
[-0.37065629, -0.36541822, -0.36015840],
[-0.41055669, 0.60004777, 0.00536958]],
dates, tickers))])
assert_output(fn, fn_inputs, fn_correct_outputs)