# Tushare + wind API 能用tushare就tushare 实在不行上windAPI 不用csmar
token = '654d36bf9bb086cb8c973e0f259e38c3efe24975386b7922e88a4cf2'
import tushare as ts
ts.set_token(token)
pro = ts.pro_api()
import utils
utils.setdir_fctr()
s_path = Path("_saved_factors")
if not os.path.exists(s_path):
os.mkdir(s_path)
# %%
# use split-adjusted share prices
Monthly_Quotation_sa = utils.cleandf(
pd.read_csv(Path('data', 'buffer', 'MonFactorPrcd_sa.csv')))
Monthly_Quotation_sa = utils.todate(Monthly_Quotation_sa,
'end_date',
format='%Y-%m-%d')
Monthly_Quotation_sa = Monthly_Quotation_sa.set_index(["ts_code", 'end_date'
]).sort_index()
Monthly_Quotation_sa['monthly_return'] = Monthly_Quotation_sa.groupby(
['ts_code'])['close'].pct_change()
# load risk free rate
rf = pd.read_csv(Path('_saved_factors', 'MacroFactor.csv'),
index_col=0,
parse_dates=['end_date'])[['RiskFreeRate']]
rf["Mon_rfr"] = (1 + rf['RiskFreeRate'] / 100)**(1 / 12) - 1
# rf = rf.sort_index().shift(1) # shift by 1
token = '654d36bf9bb086cb8c973e0f259e38c3efe24975386b7922e88a4cf2'
import tushare as ts
ts.set_token(token)
pro = ts.pro_api()
_paths = os.getcwd().split('/')
if _paths[-1] == "code":
os.chdir("..")
# %%
'''
Quarter_data
去重、查看
'''
from utils import cleandf
Quarter_data = cleandf(pd.read_csv(Path('data', 'buffer', 'QuarterFactorRaw.csv')))
Quarter_data = Quarter_data.sort_values(by=['ts_code','end_date'],ascending=[True,True])
Quarter_data = Quarter_data[~(Quarter_data.end_date.isna())]
Quarter_data = Quarter_data.drop_duplicates()
Quarter_data['end_date'] = Quarter_data['end_date'].astype(int)
date = [datetime.strptime(str(i), "%Y%m%d") for i in Quarter_data.end_date.values]
Quarter_data.loc[:, 'end_date'] = date
Quarter_data = Quarter_data.reindex(range(len(Quarter_data)))
#qgrid.show_grid(Quarter_data.loc[:,['ts_code','end_date','ann_date']])
#所有差分、roll的函数加一个.groupby，如果是if Quarter_data['is_beginning'] = 1, 则自动为Nan
Quarter_data['is_beginning'] = 0
Quarter_data = Quarter_data.sort_values(['ts_code', 'end_date'])
pro = ts.pro_api()
_paths = os.getcwd().split('/')
if _paths[-1] == "code":
os.chdir("..")
s_path = Path("_saved_factors")
if not os.path.exists(s_path):
os.mkdir(s_path)
import utils
#%%
# load data
_load = False
if _load:
Daily_Quotation = utils.cleandf(pd.read_csv(Path('data', 'buffer', 'DayFactorPrcd.csv')))
Daily_Quotation = Daily_Quotation.rename(columns={'代码': 'ts_code', '日期': 'trade_date', '成交量(股)': 'volume'})
Daily_Quotation['trade_date'] = pd.to_datetime(Daily_Quotation['trade_date'])
Daily_Quotation['end_date'] = Daily_Quotation['trade_date'] + pd.offsets.MonthEnd(0)
Daily_Quotation.index = range(len(Daily_Quotation))
Daily_Quotation = Daily_Quotation.sort_values(by=['ts_code', 'trade_date'])
Daily_Quotation.to_pickle(Path('data', 'buffer', 'DayFactorPrcd.pkl'))
else:
Daily_Quotation = pd.read_pickle(Path('data', 'buffer', 'DayFactorPrcd.pkl'))
Daily_vol = Daily_Quotation[['ts_code', 'trade_date', 'volume']] #数据原始太大了 拆分做
# %%
'''
3.aeavol -Quarterly:
Average daily trading volume(vol) for 3 days around(?before) earnings announcement -
_paths = os.getcwd().split('/')
if _paths[-1] == "code":
os.chdir("..")
s_path = Path("_saved_factors")
if not os.path.exists(s_path):
os.mkdir(s_path)
import utils
#%%
# load data
_load = False
if _load:
Weekly_Quotation = utils.cleandf(
pd.read_csv(Path('data', 'buffer', 'WeekFactorPrcd.csv'),
encoding='gbk'))
Weekly_Quotation = Weekly_Quotation.rename(
{
'代码': 'ts_code',
'日期': 'trade_date'
}, axis=1)
Weekly_Quotation = Weekly_Quotation.sort_values(
by=['ts_code', 'trade_date'])
Weekly_Quotation.index = range(len(Weekly_Quotation))
Weekly_Quotation.to_pickle(Path('data', 'buffer', 'WeekFactorPrcd.pkl'))
else:
Weekly_Quotation = pd.read_pickle(
Path('data', 'buffer', 'WeekFactorPrcd.pkl'))
# %%
'''
f27_dir = Path('data', 'divi')
f27_pt = Path('data', 'factor27_divi.csv')
if not os.path.exists(f27_pt):
'''下载divi数据'''
from utils import load_divi
load_divi(f27_dir)
dfs = pd.DataFrame()
for f in tqdm(os.listdir(f27_dir)):
divi = pd.read_csv(f27_dir / f)
dfs = pd.concat([dfs, divi], axis=0)
dfs = cleandf(dfs)
dfs.to_csv(f27_pt, index=False)
# %%
divi = pd.read_csv(f27_pt)
divi['end_date'] = divi['end_date'].astype(int).astype(str)
divi['end_date'] = pd.to_datetime(divi['end_date']) + pd.offsets.QuarterEnd(0)
divi = divi.drop_duplicates(['ts_code', 'end_date'],
keep='last') # must drop duplicates
2
divi['pay_01'] = divi['pay_date'] * 0 + 1
divi['pay_01'] = divi['pay_01'].fillna(value=0)
divi = divi.sort_values(by=['ts_code', 'end_date'], ascending=[True, True])
divi['raw_divi'] = divi.groupby(['ts_code'])['pay_01'].diff().fillna(0)
import tushare as ts
ts.set_token(token)
pro = ts.pro_api()
import utils
utils.setdir_fctr()
s_path = Path("_saved_factors")
if not os.path.exists(s_path):
os.mkdir(s_path)
# %%
# load data
_load = True
if _load:
Monthly_Quotation = utils.cleandf(
pd.read_csv(Path('data', 'buffer', 'MonFactorPrcd.csv')))
Monthly_Quotation = utils.todate(Monthly_Quotation,
'end_date',
format='%Y-%m-%d')
Monthly_Quotation.index = range(len(Monthly_Quotation))
Monthly_Quotation.to_pickle(Path('data', 'buffer', 'MonFactorPrcd.pkl'))
else:
Monthly_Quotation = pd.read_pickle(
Path('data', 'buffer', 'MonFactorPrcd.pkl'))
Ind_fctr = pd.get_dummies(Monthly_Quotation.set_index(['ts_code', 'end_date'
])['industry'],
prefix='Ind').reset_index()
Ind_fctr.to_csv(Path('_saved_factors', 'IndFactor.csv'), index=False)
# %%
'''