def __init__(self, *args, **kwargs):
self._logger = logging.getLogger("jpbank.quants")
self._logger.info("{0} initializing...".format(self.__class__.__name__))
self._start_date = kwargs.get('start_date', date(2001, 1, 1))
self._end_date = kwargs.get('end_date', date.today())
self._ar_diff = kwargs.get('ar_diff', 3)
self._frequency = kwargs.get('frequency', 'weekly')
if self._frequency == 'daily':
self._date_list = cf.create_daily_datelist(self._start_date, self._end_date)
self._term = int(kwargs.get('term_year', 3) * 261)
elif self._frequency == 'weekly':
self._date_list = cf.create_weekly_datelist(self._start_date, self._end_date)
self._term = int(kwargs.get('term_year', 3) * 52)
else:
self._date_list = cf.create_monthly_datelist(self._start_date, self._end_date)
self._term = int(kwargs.get('term_year', 3) * 12)
self._ccy_list = kwargs.get('ccy_list',
np.sort(['USDJPY Index', 'EURJPY Index', 'AUDJPY Index', 'GBPJPY Index',
'CADJPY Index', 'CHFJPY Index', 'NZDJPY Index']).tolist())
self._order = kwargs.get('order', 0)
if self._order == -1:
self._reg = 'nc'
elif self._order == 0:
self._reg = 'c'
elif self._oder == 1:
self._reg = 'ct'
self._fx_rate_df = np.log(self._get_fx_rate())
self._logger.info("{0} initialized.".format(self.__class__.__name__))
def _create_return_df(self, target_day=None):
if self._is_weekly:
date_list = cf.create_weekly_datelist(self._start_date, self._end_date)
else:
date_list = cf.create_monthly_datelist(self._start_date, self._end_date)
weekly_price = self._price_df.loc[date_list]
if target_day is None:
return pd.DataFrame(np.log(np.array(weekly_price.iloc[1:]) \
/ np.array(weekly_price.iloc[:-1])),
index = weekly_price.index[:-1],
columns = ['Return'])
else:
target_date_list = cf.create_weekly_datelist(self._start_date,
self._end_date,
weeknum=target_day)
target_price = self._price_df.loc[target_date_list]
if target_date_list[0] > date_list[0]:
if target_price.shape[0] < weekly_price.shape[0]:
weekly_price = weekly_price.iloc[:-1]
elif target_price.shape[0] == weekly_price.shape[0]:
target_price = target_price.iloc[1:]
weekly_price = weekly_price.iloc[:-1]
elif target_date_list[0] < date_list[0]:
if target_price.shape[0] > weekly_price.shape[0]:
target_price = target_price.iloc[1:]
elif target_price.shape[0] == weekly_price.shape[0]:
target_price = target_price.iloc[1:]
weekly_price = weekly_price.iloc[:-1]
return pd.DataFrame(np.log(np.array(target_price) \
/ np.array(weekly_price)),
index = weekly_price.index,
columns = ['Return'])
def __init__(self, *args, **kwargs):
self._start_date = kwargs.get('StartDate', date(2000,1,1))
self._end_date = kwargs.get('EndDate', date(2020, 4 , 17))
self._target_ccy = kwargs.get('TargetCcy', ['ZAR', 'USD'])
self._is_weekly = kwargs.get('is_weekly', True)
self._label_tickers = kwargs.get('label_tickers', ['NFCIINDX Index', 'GSUSFCI Index'])
threshold_list = kwargs.get('threshold', [5, 0.6])
assert len(threshold_list) == len(self._label_tickers)
self._threshold_dic = {}
for i in range(len(self._label_tickers)):
self._threshold_dic[self._label_tickers[i]] = threshold_list[i]
self._price_ticker = [self._target_ccy[1] + self._target_ccy[0] + ' Index']
self._price_df = self.create_factor(self._price_ticker)
if self._is_weekly:
self._date_list = cf.create_weekly_datelist(self._start_date, self._end_date)
else:
self._date_list = cf.create_monthly_datelist(self._start_date, self._end_date)
self._surprise_ticker = 'CESI' + self._target_ccy[0] + ' Index'
self._datachange_ticker = 'CECIC' + self._target_ccy[0] + ' Index'
self._ctot_ticker = 'CTOT' + self._target_ccy[0] + ' Index'
self._value_ticker = ['BISB' + ccy[:2] + 'N Index' for ccy in self._target_ccy]
#self._value_ticker = 'CTTWBR' + self._target_ccy[0][:2] + 'N Index'
self._carry_ticker_dic = {'USD':'USGG2YR Index',
'ZAR':'GSAB2YR Index',
'MXN':'GMXN02YR Index',
'TRY':'GTRU2YR Index'}
def _create_carry_df(self, col_name='Carry'):
if self._is_weekly:
date_list = cf.create_weekly_datelist(self._start_date, self._end_date)
else:
date_list = cf.create_monthly_datelist(self._start_date, self._end_date)
carry_df = self.create_factor(self._carry_ticker)
carry_df[col_name] = carry_df[self._carry_ticker[0]]\
-carry_df[self._carry_ticker[1]]
return carry_df[[col_name]].loc[date_list]
def _create_iv_df(self, vi_type='R', delta=25, vi_term='1W'):
if self._is_weekly:
date_list = cf.create_weekly_datelist(self._start_date, self._end_date)
else:
date_list = cf.create_monthly_datelist(self._start_date, self._end_date)
target_ticker = self._target_ccy[1] + self._target_ccy[0] \
+ str(delta) + vi_type + vi_term + ' BGN Curncy'
return self.create_factor([target_ticker]).loc[date_list]
def _create_rsi(self, rsi_param=7, col_name='RSI'):
if self._is_weekly:
date_list = cf.create_weekly_datelist(self._start_date, self._end_date)
else:
date_list = cf.create_monthly_datelist(self._start_date, self._end_date)
return pd.DataFrame([ta.RSI(np.array(self._price_df.iloc[:, i]), rsi_param)
for i in range(self._price_df.shape[1])],
index=[col_name],
columns=self._price_df.index).T.loc[date_list]
def _create_ctot_df(self, col_name='CTOT'):
if self._is_weekly:
date_list = cf.create_weekly_datelist(self._start_date, self._end_date)
else:
date_list = cf.create_monthly_datelist(self._start_date, self._end_date)
ctot_df = self.create_factor(self._ctot_ticker)
ctot_df[col_name] = ctot_df['CTOT'+self._target_ccy[1]+' Index']\
-ctot_df['CTOT'+self._target_ccy[0]+' Index']
return ctot_df[[col_name]].loc[date_list]
def _create_value_df(self, col_name='Value'):
if self._is_weekly:
date_list = cf.create_weekly_datelist(self._start_date, self._end_date)
else:
date_list = cf.create_monthly_datelist(self._start_date, self._end_date)
value_df = self.create_factor(self._value_ticker)
value_df[col_name] = value_df['BISB'+self._target_ccy[1][:2]+'N Index']\
-value_df['BISB'+self._target_ccy[0][:2]+'N Index']
return value_df[[col_name]].loc[date_list]
def _create_surprise_df(self, col_name='Surprise'):
if self._is_weekly:
date_list = cf.create_weekly_datelist(self._start_date, self._end_date)
else:
date_list = cf.create_monthly_datelist(self._start_date, self._end_date)
surprise_df = self.create_factor(self._surprise_ticker)
surprise_df[col_name] = surprise_df['CESI'+self._target_ccy[1]+' Index'] \
- surprise_df['CESI'+self._target_ccy[0]+' Index']
return surprise_df[[col_name]].loc[date_list]
def _create_datachange_df(self, col_name='DataChange'):
if self._is_weekly:
date_list = cf.create_weekly_datelist(self._start_date, self._end_date)
else:
date_list = cf.create_monthly_datelist(self._start_date, self._end_date)
datachange_df = self.create_factor(self._datachange_ticker)
if 'JPY' in self._price_ticker[0]:
datachange_df[col_name] = datachange_df['CECIC'+self._target_ccy[1]+' Index']
else:
datachange_df[col_name] = datachange_df['CECIC'+self._target_ccy[1]+' Index']\
-datachange_df['CECIC'+self._target_ccy[0]+' Index']
return datachange_df[[col_name]].loc[date_list]
def _calc_aggregated_hitratio(self, return_df, is_monthly=True):
if is_monthly:
date_list = cf.create_monthly_datelist(
start_date=return_df.index[0], end_date=return_df.index[-1])
else:
date_list = cf.create_weekly_datelist(
start_date=return_df.index[0], end_date=return_df.index[-1])
date_list = np.array(date_list)
aggr_return = copy.deepcopy(return_df)
aggr_return['AggregateDate'] = [date_list[date_list>=return_df.index[i]][0]
if np.any(date_list>=return_df.index[i]) \
else return_df.index[-1]\
for i in range(return_df.shape[0])]
grouped_df = aggr_return.groupby(['AggregateDate']).sum()
return pd.DataFrame(
grouped_df.where(grouped_df > 0).count() / grouped_df.shape[0]).T
def _create_label_df(self):
if self._is_weekly:
date_list = cf.create_weekly_datelist(self._start_date, self._end_date)
else:
date_list = cf.create_monthly_datelist(self._start_date, self._end_date)
label_query = "SELECT ValueDate, Ticker, Last \
FROM bbg_marketprice \
WHERE ValueDate BETWEEN '{0}' AND '{1}' \
AND Ticker IN ('{2}')".format(self._start_date, self._end_date,
"','".join(self._label_tickers))
with util.DBConnector(DBName='marketdb') as db_conn:
label_src_df = db_conn.get_data(label_query).pivot(index='ValueDate',
columns='Ticker',
values='Last').dropna(axis=0)
label_src_df['NFCIINDX Index'] = [np.nan] + label_src_df['NFCIINDX Index'].iloc[:-1].tolist()
label_src_df.dropna(axis=0, inplace=True)
#import pdb;pdb.set_trace()
for ticker in self._label_tickers:
label_src_df[ticker] = self._normalize_data(label_src_df[[ticker]], term=104)
#import pdb;pdb.set_trace()
#label_src_df[ticker] = label_src_df[ticker].apply(lambda x: x if abs(x) < self._threshold_dic[ticker] else 0)
import pdb;pdb.set_trace()
label_src_df.to_csv('fc_label_src.csv')
label_list = []
for i in range(label_src_df.shape[0]):
if label_src_df.iloc[i].max() < 0:
label = [0, 0, 1]
elif label_src_df.iloc[i][self._label_tickers[0]] > label_src_df.iloc[i][self._label_tickers[1]]:
label = [1, 0, 0]
else:
label = [0, 1, 0]
label_list.append(label)
return pd.DataFrame(label_list,
index=label_src_df.index,
columns=self._label_tickers + ['N/A']).reindex(date_list)
def __init__(self, **kwargs):
super().__init__(**kwargs)
self._logger.info("{0} initializing...".format(
self.__class__.__name__))
self._start_date = kwargs.get('start_date', None)
self._end_date = kwargs.get('end_date', None)
interval = kwargs.get('interval', 1)
frequency = kwargs.get('frequency', 'weekly')
if frequency == 'daily':
date_list = cf.create_daily_datelist(self._start_date,
self._end_date)
elif frequency == 'weekly':
date_list = cf.create_weekly_datelist(self._start_date,
self._end_date)
else:
date_list = cf.create_monthly_datelist(self._start_date,
self._end_date)
coint_vec_file = kwargs.get('coint_Vec_file', './input/coint_vec.csv')
self._weight_df = cf.convert_date_format(pd.read_csv(coint_vec_file))\
.set_index('ValueDate').loc[date_list]
self._fx_rate_df = np.log(
cf.get_fx_rate(self._start_date, self._end_date,
self._weight_df.columns.tolist())).loc[date_list]
#import pdb;pdb.set_trace()
self._coint_index_df = pd.DataFrame(
(self._fx_rate_df *
self._weight_df[self._fx_rate_df.columns]).sum(axis=1),
columns=['Price']).loc[date_list]
#self._index_return_df = pd.DataFrame(self._coint_index_df[interval:] - self._coint_index_df[:-interval],
# index = self._fx_rate_df.index[:-interval],
# columns = ['Return'])#.loc[date_list]
self._logger.info("{0} initialized.".format(self.__class__.__name__))
