def bp(self):
pb_daily = self.pb_daily
pb = CALFUNC.d_freq_to_m_freq(pb_daily)
bp = 1 / pb
res = CALFUNC.del_dat_early_than(bp, START_YEAR)
return res
def ep(self):
pe_daily = self.pe_daily
pe = CALFUNC.d_freq_to_m_freq(pe_daily)
ep = 1 / pe
res = CALFUNC.del_dat_early_than(ep, START_YEAR)
return res
def MIDCAP(self):
lncap = np.log(self.negotiablemv_daily * 10000)
lncap = CALFUNC.d_freq_to_m_freq(lncap)
y = lncap**3
X = lncap
y = y.T
X = X.T
resid = pd.DataFrame()
for code in y.columns:
y_ = y[[code]]
x_ = X[[code]]
x_['const'] = 1
dat = pd.concat([x_, y_], axis=1)
dat = dat.dropna(how='any', axis=0)
X_, y_ = dat.iloc[:, :-1], dat.iloc[:, -1:]
if len(y_) > 0:
model = sm.WLS(y_, X_)
result = model.fit()
params_ = result.params
resid_ = y_ - pd.DataFrame(
np.dot(X_, params_), index=y_.index, columns=[code])
else:
resid_ = pd.DataFrame([np.nan] * len(y),
index=y.index,
columns=[code])
resid = pd.concat([resid, resid_], axis=1)
resid = resid.T
resid = CALFUNC.del_dat_early_than(resid, START_YEAR)
return resid
def return_n_m(self):
close = self.closeprice_daily
adj = self.adjfactor
close, adj = self._align(close, adj)
c_p = close * adj
c_p = c_p.T
c_v = c_p.values
hh, ll = c_v.shape
# 1个月、3个月、6个月、12个月
m1 = np.zeros(c_v.shape)
m3 = np.zeros(c_v.shape)
m6 = np.zeros(c_v.shape)
m12 = np.zeros(c_v.shape)
for i in range(21, ll):
m1[:, i] = c_v[:, i] / c_v[:, i - 21]
for i in range(21 * 3, ll):
m3[:, i] = c_v[:, i] / c_v[:, i - 21 * 3]
for i in range(21 * 6, ll):
m6[:, i] = c_v[:, i] / c_v[:, i - 21 * 6]
for i in range(21 * 12, ll):
m12[:, i] = c_v[:, i] / c_v[:, i - 21 * 12]
m1_df = pd.DataFrame(data=m1, index=c_p.index, columns=c_p.columns)
m3_df = pd.DataFrame(data=m3, index=c_p.index, columns=c_p.columns)
m6_df = pd.DataFrame(data=m6, index=c_p.index, columns=c_p.columns)
m12_df = pd.DataFrame(data=m12, index=c_p.index, columns=c_p.columns)
m1_df_m = CALFUNC.d_freq_to_m_freq(m1_df)
m3_df_m = CALFUNC.d_freq_to_m_freq(m3_df)
m6_df_m = CALFUNC.d_freq_to_m_freq(m6_df)
m12_df_m = CALFUNC.d_freq_to_m_freq(m12_df)
m1_df_m1 = CALFUNC.del_dat_early_than(m1_df_m, START_YEAR)
m3_df_m1 = CALFUNC.del_dat_early_than(m3_df_m, START_YEAR)
m6_df_m1 = CALFUNC.del_dat_early_than(m6_df_m, START_YEAR)
m12_df_m1 = CALFUNC.del_dat_early_than(m12_df_m, START_YEAR)
res_dict = {
'RETURN_1M': m1_df_m1 - 1,
'RETURN_3M': m3_df_m1 - 1,
'RETURN_6M': m6_df_m1 - 1,
'RETURN_12M': m12_df_m1 - 1,
}
return res_dict
def is_open(self):
open = self.openPrice_daily
high = self.highprice_daily
low = self.lowPrice_daily
if self._status == 'all':
# 不是停牌的
is_open = ~pd.isna(open)
# 不是开盘涨跌停的
tmp1 = open == high
tmp2 = high == low
tmp = ~(tmp1 & tmp2)
is_open = tmp & is_open
is_open = CALFUNC.d_freq_to_m_freq(is_open, shift=True)
is_open = CALFUNC.del_dat_early_than(is_open, START_YEAR)
elif self._status == 'update':
factor = self.IS_OPEN
# 先删除过去计算的bug
to_del = [c for c in factor.columns if c not in self._mes]
factor.drop(to_del, axis=1, inplace=True)
latest_dt = factor.columns[-1]
# 删除无用的日频数据
saved_cols = [i for i in open.columns if i > latest_dt]
open = open[saved_cols]
high = high[saved_cols]
low = low[saved_cols]
is_open = ~pd.isna(open)
# 不是开盘涨跌停的
tmp1 = open == high
tmp2 = high == low
tmp = ~(tmp1 & tmp2)
is_open = tmp & is_open
is_open = CALFUNC.d_freq_to_m_freq(is_open, shift=True)
is_open = pd.concat([factor, is_open], axis=1)
return is_open
def LNCAP_Barra(self):
lncap = np.log(self.negotiablemv_daily * 10000)
lncap = CALFUNC.d_freq_to_m_freq(lncap)
lncap = CALFUNC.del_dat_early_than(lncap, START_YEAR)
return lncap
def Mkt_cap_float(self):
negotiablemv = self.negotiablemv_daily
negotiablemv = CALFUNC.d_freq_to_m_freq(negotiablemv)
res = CALFUNC.del_dat_early_than(negotiablemv, START_YEAR)
return res
