def __init__(self, start, flag, fre, factor_filename, day_range, dirpath, date=None):
self.sql = "select s_info_windcode,trade_dt,s_dq_freeturnover from AShareEODDerivativeIndicator"
self.flag = flag # 0 回测 1更新
self.day_range = day_range ##
self.dirpath = dirpath # 新建data文件夹 存各种因子数据
self.factor_filename = factor_filename # data文件夹里 具体因子的文件夹
self.factor_path = self.dirpath + '/' + self.factor_filename
# 确定运行日期
if date == None:
self.today = strftime("%Y%m%d", localtime())
else:
self.today = date
self.initial = '20080101' # 留点窗口 为了下面start往前推能取到日期
self.start = start
self.fre = fre
self.all_trade_day = get_tradeDay.wind(self.initial, self.today, fre='day') ##用于定位一段交易日区间
self.trade_day = get_tradeDay.wind(self.start, self.today, fre=self.fre)
# self.trade_day = self.trade_day.iloc[:-1]
self.getdb = client_db.read_db(type='ctquant2')
if not os.path.exists(self.dirpath):
os.mkdir(self.dirpath)
if not os.path.exists('%s' % (self.factor_path)):
os.mkdir('%s' % (self.factor_path))
def run(output_path, start, end, flag, fre, dir_path):
factor_name1 = [
'Style.SIZE', 'Style.RESVOL', 'Style.SIZENL', 'Style.LIQUIDTY'
]
factor_name = [
'SIZE_barra', 'RESVOL_barra', 'SIZENL_barra', 'LIQUIDTY_barra'
]
for i in factor_name:
if not os.path.exists(output_path + '/' + i):
os.mkdir(output_path + '/' + i)
trade_day = get_tradeDay.wind(start, end, fre=fre)
if flag == 0:
for i in range(len(trade_day)):
today = trade_day.iloc[i]
save_day = today
print(save_day)
update(save_day, factor_name, factor_name1, dir_path, output_path)
elif flag == 1:
save_day = end
print(save_day)
update(save_day, factor_name, factor_name1, dir_path, output_path)
def run(firstRun_update, start, end, fre, index_code, dir_path,
beachmark_file):
if not os.path.exists('%s' % beachmark_file):
os.mkdir('%s' % beachmark_file)
if not os.path.exists('%s/%s' % (beachmark_file, index_code)):
os.mkdir('%s/%s' % (beachmark_file, index_code))
if firstRun_update == 0:
trade_day = get_tradeDay.wind(start, end, fre)
for i in trade_day:
print(i)
df = pd.read_excel('%s/000846weightnextday%s.xls' % (dir_path, i),
dtype={u'成分券代码\nConstituent Code': str})
df = df.iloc[:, [4, 16]]
df.columns = ['code', 'weight']
df['code'] = df['code'].apply(lambda x: x + '-CN')
df.to_csv('%s/%s/benchmark_%s.csv' %
(beachmark_file, index_code, i),
index=None,
header=None)
elif firstRun_update == 1:
cal = Calendar('China.SSE')
today = strftime("%Y%m%d", localtime())
today = Date.strptime(today, '%Y%m%d')
today = cal.advanceDate(today, Period('-1b'))
today = today.strftime("%Y%m%d")
df = pd.read_excel('%s/000846weightnextday%s.xls' % (dir_path, today),
dtype={u'成分券代码\nConstituent Code': str})
df = df.iloc[:, [4, 16]]
df.columns = ['code', 'weight']
df['code'] = df['code'].apply(lambda x: x + '-CN')
df.to_csv('%s/%s/benchmark_%s.csv' %
(beachmark_file, index_code, today),
index=None,
header=None)
def get_fcftar(self):
self.trade_day = get_tradeDay.wind('20170701',
self.today,
fre=self.fre)
if not os.path.exists(self.raw_dirpath):
os.mkdir(self.raw_dirpath)
for fre in self.fre_shift_list:
file1 = self.raw_dirpath + '/' + self.factor_filename1 + '_' + self.cal_model1 + '_' + str(
fre)
file2 = self.raw_dirpath + '/' + self.factor_filename2 + '_' + self.cal_model2 + '_' + str(
fre)
new_factor_path = self.raw_dirpath + '/' + self.new_factor_name + str(
fre)
if not os.path.exists(new_factor_path):
os.mkdir(new_factor_path)
if self.flag == 0:
for i in self.trade_day:
self.update(i, file1, self.factor_filename1,
self.cal_model1, file2, self.factor_filename2,
self.cal_model2, new_factor_path)
elif self.flag == 1:
self.update(self.today, file1, self.factor_filename1,
self.cal_model1, file2, self.factor_filename2,
self.cal_model2, new_factor_path)
def run(ret_path, output_path, today):
commission = 0.0023
stamp_buy = 0.0001
stamp_sell = 0.0001
day = datetime.datetime.strptime(today, '%Y%m%d')
pre_day = day - datetime.timedelta(days=40)
day = day.strftime('%Y%m%d')
pre_day = pre_day.strftime('%Y%m%d')
tradeday = get_tradeDay.wind(pre_day, today, fre = 'day')
day20 = tradeday.iloc[-20:]
day5 = tradeday.iloc[-5:]
srisk = get_data(ret_path, today)
srisk = srisk[['Symbol','srisk']]
sret5 = get_sret_5(ret_path, day5)
close20 = get_close_20(ret_path,day20)
df_tcost = pd.merge(srisk, sret5, on='Symbol')
df_tcost = pd.merge(df_tcost, close20, on='Symbol')
df_tcost['lincoeff'] = commission + 0.5*(stamp_buy + stamp_sell) + 2.5 * df_tcost['sret'] + 0.005/df_tcost['close']
df_tcost['sqrtcoeff'] = 0.001*df_tcost['srisk']
df_tcost = df_tcost[['Symbol','lincoeff','sqrtcoeff']]
df_tcost = df_tcost.sort_values('Symbol')
df_header = pd.DataFrame([[np.nan,'GROUP','GROUP'],['name','lincoeff','sqrtcoeff'],['unit','number','number'],[np.nan,np.nan,np.nan]],columns=['Symbol','lincoeff','sqrtcoeff'])
df_final = pd.concat([df_header,df_tcost])
df_final.to_csv('%s/coef_%s.csv'%(output_path,today),index=None, header= None)
def get_delta_alpha(output_path, today):
# today = '20180108'
# output_path = '.'
index_list = ['ESG100', 'HS300', 'ZZ500']
day = datetime.datetime.strptime(today, '%Y%m%d')
pre_day = day - datetime.timedelta(days=20)
day = day.strftime('%Y%m%d')
pre_day = pre_day.strftime('%Y%m%d')
trade_day = get_tradeDay.wind(pre_day, day, fre='day')
pre_day = trade_day[trade_day < day].iloc[-1]
df1 = pd.read_csv(
'Z:/daily_data/alpha/neut/alphaALL/alphaALL_neut_CN_%s.csv' % day,
skiprows=3)
df1.columns = [
u'stock_code', u'alphaALL_T', u'alphaALL_ESG100_T',
u'alphaALL_HS300_T', u'alphaALL_ZZ500_T', u'Value_T', u'Quality_T',
u'Revision_T', u'Fndsurp_T', u'IU_T', u'Mktmmt_T', u'Insider_T',
u'lincoef', u'sqrtcoef', u'UNIVERSE'
]
df1 = df1[[
u'stock_code', u'alphaALL_ESG100_T', u'alphaALL_HS300_T',
u'alphaALL_ZZ500_T'
]]
df2 = pd.read_csv(
'Z:/daily_data/alpha/neut/alphaALL/alphaALL_neut_CN_%s.csv' % pre_day,
skiprows=3)
df2.columns = [
u'stock_code', u'alphaALL_T-1', u'alphaALL_ESG100_T-1',
u'alphaALL_HS300_T-1', u'alphaALL_ZZ500_T-1', u'Value_T-1',
u'Quality_T-1', u'Revision_T-1', u'Fndsurp_T-1', u'IU_T-1',
u'Mktmmt_T-1', u'Insider_T-1', u'lincoef', u'sqrtcoef', u'UNIVERSE'
]
df2 = df2[[
u'stock_code', u'alphaALL_ESG100_T-1', u'alphaALL_HS300_T-1',
u'alphaALL_ZZ500_T-1'
]]
df = pd.merge(df1, df2, on='stock_code')
for input1 in index_list:
df['delta_alpha_%s' %
input1] = df['alphaALL_%s_T' % input1] - df['alphaALL_%s_T-1' %
input1]
col = ['delta_alpha_%s' % i for i in index_list]
df = df[['stock_code'] + col]
df_header = pd.DataFrame(
[[np.nan] + ['GROUP'] * len(col), ['name'] + col,
['unit'] + ['number'] * len(col), [np.nan] + [np.nan] * len(col)],
columns=['stock_code'] + col)
df_final = pd.concat([df_header, df])
df_final.to_csv('%s/delta_alpha_%s.csv' % (output_path, today),
index=None,
header=None)
def run(start, end, today, dirpath, output_path, flag, fre):
getdb = read_db(type='wind')
tradeday = get_tradeDay.wind(start, end, fre=fre)
tradeday = tradeday.iloc[:-1]
# if not os.path.exists(dirpath):
# os.mkdir(dirpath)
if flag == 0:
for day in tradeday:
print(day)
get_holding(day, dirpath, output_path, getdb)
else:
get_holding(today, dirpath, output_path, getdb)
def run(start, end, today, dirpath, flag, fre):
getdb = client_db.read_db()
tradeday = get_tradeDay.wind(start, end, fre=fre)
tradeday = tradeday.iloc[:-1]
if not os.path.exists(dirpath):
os.mkdir(dirpath)
if flag == 0:
for day in tradeday:
print(day)
suspend(day, dirpath, getdb)
else:
suspend(today, dirpath, getdb)
def __init__(self, start, end, fre, factor_name_list, type, benchmark,
input_path, universe_path, n_quantile, outputpath):
self.start = start
self.end = end
self.fre = fre
self.factor_name_list = factor_name_list
self.tradeday = get_tradeDay.wind(self.start, self.end, fre=self.fre)
self.daily = get_tradeDay.wind(self.start, self.end, fre='day')
self.type = type
self.benchmark = benchmark
self.input_path = input_path
self.universe_path = universe_path
self.n_quantile = n_quantile
self.pct_quantiles = 1 / float(self.n_quantile)
self.outputpath0 = outputpath
self.getdb = client_db.read_db(type='wind')
def __init__(self,
start,
sql,
flag,
factor_filename,
day_range,
date=None):
self.sql = sql
self.flag = flag # 0 回测 1更新
self.day_range = day_range ##
self.dirpath = 'raw_data' # 新建data文件夹 存各种因子数据
self.factor_filename = factor_filename # data文件夹里 具体因子的文件夹
self.factor_path = self.dirpath + '/' + self.factor_filename
# 确定运行日期
if date == None:
self.today = strftime("%Y%m%d", localtime())
else:
self.today = date
self.initial = '20170101' # 留点窗口 为了下面start往前推能取到日期
self.start = start
self.all_trade_day = get_tradeDay.wind(self.initial,
self.today,
fre='day') ##用于定位一段交易日区间
self.trade_day = get_tradeDay.wind(self.start, self.today, fre='day')
# self.trade_day = self.trade_day.iloc[:-1]
self.getdb = client_db.read_db(type='ctquant2')
if not os.path.exists(self.dirpath):
os.mkdir(self.dirpath)
if not os.path.exists('%s' % (self.factor_path)):
os.mkdir('%s' % (self.factor_path))
def event_data_prepare(self):
daily = get_tradeDay.wind('20060104', '20171109', fre='day')
daily = daily.values
df_factor = self.get_factor()
df_index_daily = self.get_index_daily()
df_close_daily = self.get_close_daily()
df_ret_daily = df_close_daily.pct_change().shift(-1)
df_index_daily['ret'] = df_index_daily['close'].pct_change()
df_index_daily = df_index_daily[['trade_day', 'ret']]
df_index_daily['ret'] = df_index_daily['ret'].shift(-1).values
df_index_daily = df_index_daily.set_index('trade_day')
df_index_daily.columns = ['benchmark']
return df_factor, daily, df_ret_daily, df_index_daily
def run(start, end, today, output_path, flag, fre):
config_path = 'Z:/daily_data/config/config.yaml'
with open(config_path) as f:
config = yaml.load(f.read())
account = config['account']
for key, sub_account in account.items():
if sub_account['type'] == 'p':
account_name = sub_account['subaccount'].split('account-')[1]
getdb = read_db(type='wind')
tradeday = get_tradeDay.wind(start, end, fre=fre)
tradeday = tradeday.iloc[:-1]
if flag == 0:
for day in tradeday:
print(day)
get_new_stock(day, account_name, output_path, getdb)
else:
get_new_stock(today, account_name, output_path, getdb)
def run(start, end, flag, dirpath, output1, output2):
if not os.path.exists(output1):
os.mkdir(output1)
if not os.path.exists(output2):
os.mkdir(output2)
if flag == 0:
fre = 'day'
tradeday = get_tradeDay.wind(start, end, fre=fre)
for day in tradeday:
print(day)
get_exposure(dirpath, output1, day)
get_cov(dirpath, output2, day)
else:
today = end
# today = datetime.strftime(today.toDateTime(), "%Y%m%d")
get_exposure(dirpath, output1, today)
get_cov(dirpath, output2, today)
def run(start, end, flag, dirpath, output1, output2):
if not os.path.exists(output1):
os.mkdir(output1)
if not os.path.exists(output2):
os.mkdir(output2)
if flag == 0:
fre = 'day'
tradeday = get_tradeDay.wind(start, end, fre=fre)
for day in tradeday:
print(day)
get_exposure(dirpath, output1, day)
get_cov(dirpath, output2, day)
else:
end = Date.strptime(end, '%Y%m%d')
today = cal.advanceDate(end, Period('-1b'))
today = today.strftime("%Y%m%d")
# today = datetime.strftime(today.toDateTime(), "%Y%m%d")
get_exposure(dirpath, output1, today)
get_cov(dirpath, output2, today)
start = '20080401'
end = '20170510'
df_index = DataAPI.MktIdxdGet(tradeDate=u"",
indexID=u"",
ticker=u"000300",
beginDate=start,
endDate=end,
exchangeCD=u"XSHE,XSHG",
field=['tradeDate', 'closeIndex'],
pandas="1")
df_index.columns = ['trade_day', 'close']
fre = 'month'
tradeday = get_tradeDay.wind(start, end, fre=fre)
factor_name = 'RAND1'
factor_path = 'Z:/axioma_data/alpha/%s' % factor_name
# factor_path = 'E:/QUANT/alpha_factor/risk_data/SIZE_uqer'
close_path = 'Z:/backtest_data/close_data/gogo_data'
df_close = pd.DataFrame([])
for i in tradeday:
print(i)
close = pd.read_csv('%s/close_%s.csv' % (close_path, i),
dtype={'ticker': str})
close['trade_day'] = i
df_close = pd.concat([df_close, close])
df_close = df_close.pivot(index='trade_day',
from tools import get_tradeDay
reload(sys)
start = '20080401'
end = '20170510'
df_index = DataAPI.MktIdxdGet(tradeDate=u"", indexID=u"", ticker=u"000300", beginDate=start, endDate=end,
exchangeCD=u"XSHE,XSHG", field=['tradeDate', 'closeIndex'], pandas="1")
df_index.columns = ['trade_day', 'close']
df_index['trade_day'] = df_index['trade_day'].apply(lambda x: x[:4] + x[5:7] + x[8:])
fre = 'month'
tradeday = get_tradeDay.wind(start, end, fre=fre)
daily = get_tradeDay.wind(start, end, fre='day')
factor_name = 'fcff_sign_r2'
# factor_path = 'Z:/axioma_data/alpha/%s' % factor_name
factor_path = 'E:/factor_data/filter_data/%s'%factor_name
universe_path = 'Z:/axioma_data/universe'
outputpath = 'E:/QUANT/analyse_factor/fig2'
outputpath = '%s/%s' % (outputpath, factor_name)
if not os.path.exists(outputpath):
os.mkdir(outputpath)
close_path = 'Z:/backtest_data/close_data/gogo_data'
df = pd.merge(df1, df2, on='stock_code')
for input1 in index_list:
df['delta_alpha_%s' %
input1] = df['alphaALL_%s_T' % input1] - df['alphaALL_%s_T-1' %
input1]
col = ['delta_alpha_%s' % i for i in index_list]
df = df[['stock_code'] + col]
df_header = pd.DataFrame(
[[np.nan] + ['GROUP'] * len(col), ['name'] + col,
['unit'] + ['number'] * len(col), [np.nan] + [np.nan] * len(col)],
columns=['stock_code'] + col)
df_final = pd.concat([df_header, df])
df_final.to_csv('%s/delta_alpha_%s.csv' % (output_path, today),
index=None,
header=None)
if __name__ == '__main__':
start = '20180103'
end = '20180402'
output_path = 'Z:/daily_data/tcost_model'
trade_day = get_tradeDay.wind(start, end, fre='day')
for today in trade_day:
print(today)
get_delta_alpha(output_path, today)
