def find_best_index(stock,
indices,
lookback=252,
stock_ceiling_params=DEFAULT_STOCK_CEILING_PARAMS,
index_floor_params=DEFAULT_INDEX_FLOOR_PARAMS,
best_fit_param=BEST_FIT_PERCENTILE):
"""For a given stock and group of indices, find the index that has the best fit based on OLS regression"""
scrub_params_all = [
get_scrub_params(stock, index, lookback, stock_ceiling_params,
index_floor_params, best_fit_param)
for index in indices
]
betas = [
create_beta_object_from_scrub_params(stock, indices[i], lookback,
scrub_params_all[i])
for i in range(len(indices))
]
# OLS Info
corrs = [beta.corr for beta in betas]
highest_corr = max(corrs)
highest_corr_index = corrs.index(highest_corr)
best_index = indices[highest_corr_index]
return best_index
def run():
stock = 'NBIX'
stock2 = 'NBIX'
index = 'XBI'
beta_lookback = 252
chart_lookback = beta_lookback
base = 100
scrub_params = ScrubParams(.125, .01, 90)
stock_ceiling_params = Stock_Ceiling_Params(initial_ceiling=.20, SD_multiplier=4.0)
index_floor_params = Index_Floor_Params(SD_multiplier=1.0)
scrub_params = get_scrub_params(stock,
index,
stock_ceiling_params=stock_ceiling_params,
index_floor_params=index_floor_params)
beta = Beta(stock, index, beta_lookback, scrub_params)
print(beta.num_days_in_calculation)
# print(beta_value.degrees_of_freedom)
print(beta.beta_value)
print(beta.beta1)
# beta_value.describe()
beta.show_scrub_trajectory()
# beta2 = Beta(stock2, index, beta_lookback, scrub_params).beta_value
# beta_value, beta2 = 0.0, 0.0
# Stock Lines to plot
stock_line = StockLineSimple(stock, chart_lookback, base)
index_line = StockLineSimple(index, chart_lookback, base)
stock_line_adj = StockLineBetaAdjusted(stock, chart_lookback, beta.beta_value, index, base)
stock_lines = [stock_line.stock_line(color='red'),
index_line.stock_line(color='black'),
stock_line_adj.stock_line(color='blue'),
]
StockChart(stock_lines).run()
Beta(stock, index, beta_lookback, ScrubParams(.075, .01, 95)).show_scrub_trajectory()
print('End of Module')
def best_beta_value(self):
index = 'XBI'
beta_lookback = 252
stock_ceiling_params = Stock_Ceiling_Params(initial_ceiling=.20,
SD_multiplier=4.0)
index_floor_params = Index_Floor_Params(SD_multiplier=1.0)
best_fit_param = 95
scrub_params = get_scrub_params(
self.stock,
index,
stock_ceiling_params=stock_ceiling_params,
index_floor_params=index_floor_params,
best_fit_param=best_fit_param)
beta_value = Beta(self.stock, index, beta_lookback,
scrub_params).beta_value
return beta_value
def calculate_adjusted_beta(stock,
index,
lookback,
stock_ceiling_params,
index_floor_params,
best_fit_param):
"""Calculate the adjusted beta_value measurement for the stock and index over a lookback based on the three core adjustments:
- Stock Ceiling: Scrub data points where the stock moved more than the specified threshold.
- Index Floor: Scrub data points where the index moved less than the specified threshold.
- Best Fit Param: Keep only the n-percentile best fit points in the OLS regression"""
scrub_params = get_scrub_params(stock,
index,
lookback=lookback,
stock_ceiling_params=stock_ceiling_params,
index_floor_params=index_floor_params,
best_fit_param=best_fit_param)
beta_value = Beta(stock, index, lookback, scrub_params).beta_value
return beta_value
def __init__(
self,
stock: 'str',
index: 'str',
lookback: 'int' = 252,
scrub_params: 'obj' = None,
# Optional Parameters as an alternative to entering scrub_params
stock_ceiling_params=DEFAULT_STOCK_CEILING_PARAMS,
index_floor_params=DEFAULT_INDEX_FLOOR_PARAMS,
best_fit_param=BEST_FIT_PERCENTILE):
"""The Beta object takes as parameters the stock, index, lookback, and scrub_params object.
The user can enter scrub_params OR a set of stock_ceiling_params, index_floor_params, and best_fit_param
, which map to scrub_params."""
"""Calculate the adjusted beta_value measurement for the stock and index over a lookback...
based on the three core adjustments:
- Stock Ceiling: Scrub data points where the stock moved more than the specified threshold.
- Index Floor: Scrub data points where the index moved less than the specified threshold.
- Best Fit Param: Keep only the n-percentile best fit points in the OLS regression
"""
self.stock = stock
self.index = index
self.lookback = lookback
if scrub_params is None:
self.scrub_params = get_scrub_params(
stock,
index,
lookback=252,
stock_ceiling_params=DEFAULT_STOCK_CEILING_PARAMS,
index_floor_params=DEFAULT_INDEX_FLOOR_PARAMS,
best_fit_param=BEST_FIT_PERCENTILE)
else:
self.scrub_params = scrub_params
self.price_table = PriceTable.head(
self.lookback)[[self.stock, self.index]]
self.daily_returns = daily_returns(self.price_table)
self.num_data_points = self.daily_returns.shape[0]
def get_betas_over_iterable(stock,
index,
lookback=252,
stock_ceiling_params=DEFAULT_STOCK_CEILING_PARAMS,
index_floor_params=DEFAULT_INDEX_FLOOR_PARAMS,
best_fit_param=BEST_FIT_PERCENTILE,
save_to_file=False):
stock_ceiling_params = [stock_ceiling_params]
index_floor_params = [index_floor_params]
best_fit_param = [best_fit_param]
param_combos = list(
itertools.product(stock, index, lookback, stock_ceiling_params,
index_floor_params, best_fit_param))
counter = range(len(param_combos))
stocks = [i[0] for i in param_combos]
indices = [i[1] for i in param_combos]
lookbacks = [i[2] for i in param_combos]
#scrub_params_all = [get_scrub_params(stock, index, lookback, stock_ceiling_params, index_floor_params, best_fit_param) for params in param_combos]
scrub_params_all = [get_scrub_params(*params) for params in param_combos]
scrub_param_combos = list(zip(stocks, indices, lookbacks,
scrub_params_all))
betas = [
Beta(stocks[i], indices[i], lookbacks[i], scrub_params_all[i])
for i in counter
]
# OLS Info
beta_values = [beta.beta_value for beta in betas]
corrs = [beta.corr for beta in betas]
# Scrubbing Info
stock_cutoffs = [
scrub_params.stock_cutoff for scrub_params in scrub_params_all
]
index_cutoffs = [
scrub_params.index_cutoff for scrub_params in scrub_params_all
]
percentile_cutoffs = [
scrub_params.percentile_cutoff for scrub_params in scrub_params_all
]
percent_days_in_calc = [beta.percent_days_in_calculation for beta in betas]
# Beta to SPY Info
index_betas_to_SPY = [get_ETF_beta_to_SPY(index) for index in indices]
betas_to_SPY = [index_betas_to_SPY[i] * beta_values[i] for i in counter]
# Returns Info
stock_returns = [
get_total_return(stocks[i], lookbacks[i]) for i in counter
]
index_returns = [
get_total_return(indices[i], lookbacks[i]) for i in counter
]
idio_returns = [
(1 + stock_returns[i]) / (1 + index_returns[i] * beta_values[i]) - 1
for i in counter
]
# Unadjusted OLS Info
unadjusted_betas = [
Beta(stocks[i], indices[i], lookbacks[i],
ScrubParams(False, False, False)) for i in counter
]
unadjusted_beta_values = [beta.beta_value for beta in unadjusted_betas]
unadjusted_corrs = [beta.corr for beta in unadjusted_betas]
# Prepare Information for the DataFrame in an Ordered Dictionary
table_info_dict = OrderedDict([
#Unadjusted OLS Info
('Unadj. Beta', unadjusted_beta_values),
('Unadj. Corr', unadjusted_corrs),
# OLS Info
('Beta', beta_values),
('Corr', corrs),
# Beta to SPY Info
('Index_Beta_to_SPY', index_betas_to_SPY),
('Beta_to_SPY', betas_to_SPY),
# Returns Info
('Stock_Return', stock_returns),
('Index_Return', index_returns),
('Idio_return', idio_returns),
# Scrubbing Info
('Stock_Cutoff', stock_cutoffs),
('Index_Cutoff', index_cutoffs),
('Percentile_Cutoff', percentile_cutoffs),
('Percent_Days', percent_days_in_calc)
])
parameters = [stock, index, lookback]
for parameter in parameters:
if len(parameter) > 1:
iterable = parameter
else:
pass
if iterable == stock:
not_iterable = index[0]
elif iterable == index:
not_iterable = stock[0]
elif iterable == lookback:
not_iterable = 'Stock: {}, Index {}'.format(stock[0], index[0])
else:
raise ValueError
# Create DataFrame
column_labels = table_info_dict.keys()
table_info = list(zip(*table_info_dict.values()))
#index_row = pd.Index(indices, name = 'Index')
index_row = pd.Index(iterable, name='Index')
iterables_columns = [[not_iterable], column_labels]
#iterables_columns = [[stock], column_labels]
index_column = pd.MultiIndex.from_product(iterables_columns,
names=['Stock', 'Beta_Info'])
df = pd.DataFrame(table_info, index=index_row, columns=index_column)
if save_to_file:
to_pickle_and_CSV(df, file_name)
return df
def get_betas_multiple_stocks(
stocks: 'iterable of stocks',
index: 'one index',
lookback=252,
stock_ceiling_params=DEFAULT_STOCK_CEILING_PARAMS,
index_floor_params=DEFAULT_INDEX_FLOOR_PARAMS,
best_fit_param=BEST_FIT_PERCENTILE,
save_to_file=False):
scrub_params_all = [
get_scrub_params(stock, index, lookback, stock_ceiling_params,
index_floor_params, best_fit_param)
for stock in stocks
]
#betas = [create_beta_object_from_scrub_params(stocks[i], index, lookback, scrub_params_all[i]) for i in range(len(stocks))]
betas = [
Beta(stocks[i], index, lookback, scrub_params_all[i])
for i in range(len(stocks))
]
beta_values = [beta.beta_value for beta in betas]
# Returns Info
returns = [get_total_return(stock, lookback) for stock in stocks]
index_returns = [
get_total_return(index, lookback) for _ in range(len(stocks))
]
idio_returns = [
(1 + returns[i]) / (1 + index_returns[i] * beta_values[i]) - 1
for i in range(len(stocks))
]
# Beta to SPY Info
index_betas_to_SPY = [get_ETF_beta_to_SPY(stock) for stock in stocks]
betas_to_SPY = [
index_betas_to_SPY[i] * beta_values[i] for i in range(len(stocks))
]
# Prepare Information for the DataFrame in an Ordered Dictionary
info = OrderedDict([
# Index Symbol
('Index', [index for _ in range(len(stocks))]),
# OLS Info
('Beta', beta_values),
('Corr', [beta.corr for beta in betas]),
# Beta to SPY Info
('Index_Beta_to_SPY', index_betas_to_SPY),
('Beta_to_SPY', betas_to_SPY),
# Returns Info
('Stock_Return', returns),
('Index_Return', index_returns),
('Idio_Return', idio_returns),
# Scrubbing Info
('Stock_Cutoff',
[scrub_params.stock_cutoff for scrub_params in scrub_params_all]),
('Index_Cutoff',
[scrub_params.index_cutoff for scrub_params in scrub_params_all]),
('Percentile_Cutoff',
[scrub_params.percentile_cutoff
for scrub_params in scrub_params_all]),
('Percent_Days', [beta.percent_days_in_calculation for beta in betas])
])
# Create DataFrame
column_labels = info.keys()
table_info = list(zip(info.values()))
index_row = pd.Index(stocks, name='Stock')
#iterables_columns = [[index], column_labels]
iterables_columns = [['Index'], column_labels]
index_column = pd.MultiIndex.from_product(iterables_columns,
names=['Index', 'Beta_Info'])
df = pd.DataFrame(table_info, index=index_row, columns=index_column)
if save_to_file:
to_pickle_and_CSV(df, file_name)
return df
def get_betas_multiple_indices(
stock,
indices: 'iterable of indices',
lookback=252,
stock_ceiling_params=DEFAULT_STOCK_CEILING_PARAMS,
index_floor_params=DEFAULT_INDEX_FLOOR_PARAMS,
best_fit_param=BEST_FIT_PERCENTILE,
save_to_file=False):
scrub_params_all = [
get_scrub_params(stock, index, lookback, stock_ceiling_params,
index_floor_params, best_fit_param)
for index in indices
]
betas = [
Beta(stock, indices[i], lookback, scrub_params_all[i])
for i in range(len(indices))
]
# OLS Info
beta_values = [beta.beta_value for beta in betas]
corrs = [beta.corr for beta in betas]
# Scrubbing Info
stock_cutoffs = [
scrub_params.stock_cutoff for scrub_params in scrub_params_all
]
index_cutoffs = [
scrub_params.index_cutoff for scrub_params in scrub_params_all
]
percentile_cutoffs = [
scrub_params.percentile_cutoff for scrub_params in scrub_params_all
]
percent_days_in_calc = [beta.percent_days_in_calculation for beta in betas]
# Beta to SPY Info
index_betas_to_SPY = [get_ETF_beta_to_SPY(index) for index in indices]
betas_to_SPY = [
index_betas_to_SPY[i] * beta_values[i] for i in range(len(indices))
]
# Returns Info
returns = [get_total_return(stock, lookback) for _ in indices]
index_returns = [get_total_return(index, lookback) for index in indices]
idio_returns = [
(1 + returns[i]) / (1 + index_returns[i] * beta_values[i]) - 1
for i in range(len(indices))
]
# Unadjusted OLS Info
unadjusted_betas = [
Beta(stock, indices[i], lookback, ScrubParams(False, False, False))
for i in range(len(indices))
]
unadjusted_beta_values = [beta.beta_value for beta in unadjusted_betas]
unadjusted_corrs = [beta.corr for beta in unadjusted_betas]
# Prepare Information for the DataFrame in an Ordered Dictionary
table_info_dict = OrderedDict([
#Unadjusted OLS Info
('Unadj. Beta', unadjusted_beta_values),
('Unadj. Corr', unadjusted_corrs),
# OLS Info
('Beta', beta_values),
('Corr', corrs),
# Beta to SPY Info
('Index_Beta_to_SPY', index_betas_to_SPY),
('Beta_to_SPY', betas_to_SPY),
# Returns Info
('Stock_Return', returns),
('Index_Return', index_returns),
('Idio_return', idio_returns),
# Scrubbing Info
('Stock_Cutoff', stock_cutoffs),
('Index_Cutoff', index_cutoffs),
('Percentile_Cutoff', percentile_cutoffs),
('Percent_Days', percent_days_in_calc)
])
# Create DataFrame
column_labels = table_info_dict.keys()
table_info = list(zip(*table_info_dict.values()))
index_row = pd.Index(indices, name='Index')
iterables_columns = [[stock], column_labels]
index_column = pd.MultiIndex.from_product(iterables_columns,
names=['Stock', 'Beta_Info'])
df = pd.DataFrame(table_info, index=index_row, columns=index_column)
if save_to_file:
to_pickle_and_CSV(df, file_name)
return df
