def bias_turn_pnd(turnratio, sector_df, n):
bias_turnratio = bt.AZ_Rolling(turnratio, n).mean() / bt.AZ_Rolling(
turnratio, 480).mean() - 1
bias_turnratio_up = (bias_turnratio > 0).astype(int)
bias_turnratio_dn = (bias_turnratio < 0).astype(int)
target_df = bias_turnratio_up - bias_turnratio_dn
return target_df * sector_df
def CCI(high, low, close, sector_df, n, limit_list):
PP = (high + low + close) / 3
CCI_signal = (PP - bt.AZ_Rolling(PP, n).mean()) / bt.AZ_Rolling(
PP, n).std()
all_target_df = pd.DataFrame()
for limit in limit_list:
CCI_up = (CCI_signal >= limit).astype(int)
CCI_dn = -(CCI_signal <= -limit).astype(int)
CCI = CCI_up + CCI_dn
all_target_df = all_target_df.add(CCI, fill_value=0)
return all_target_df * sector_df
def pnd_vol_continue_ud(self, adj_r, sector_df, n):
vol_df = bt.AZ_Rolling(adj_r, n).std() * (250**0.5)
vol_df[vol_df < 0.08] = 0.08
vol_continue_ud_df = self.pnd_continue_ud(vol_df,
sector_df,
n_list=[3, 4, 5])
return vol_continue_ud_df
def pnd_volitality_and_more_(self, n_list):
for n in n_list:
vol_df = bt.AZ_Rolling(self.sector_adj_r, n).std() * (250**0.5)
vol_df[vol_df < 0.08] = 0.08
evol_df = bt.AZ_Rolling(vol_df, 30).apply(
lambda x: 1 if x[-1] > 2 * x.mean() else 0) * self.sector_df
vol_continue_ud_df = self.pnd_continue_ud(vol_df,
self.sector_df,
n_list=[3, 4, 5])
# 剔除极值
vol_df_count_down = 1 / (vol_df.replace(0, np.nan) *
self.sector_df)
file_name = f'vol_p{n}d'
fun = 'funda_data_deal.BaseDeal.pnd_volitality'
raw_data_path = ('EM_Funda/DERIVED_14/aadj_r.csv', )
args = (n, )
self.judge_save_fun(vol_df,
file_name,
self.save_root_path,
fun,
raw_data_path,
args,
if_filter=False)
file_name = f'vol_count_down_p{n}d'
fun = 'funda_data_deal.BaseDeal.pnd_volitality_count_down'
raw_data_path = ('EM_Funda/DERIVED_14/aadj_r.csv', )
args = (n, )
self.judge_save_fun(vol_df_count_down, file_name,
self.save_root_path, fun, raw_data_path, args)
file_name = f'evol_p{n}d'
fun = 'funda_data_deal.BaseDeal.pnd_evol'
raw_data_path = ('EM_Funda/DERIVED_14/aadj_r.csv', )
args = (n, )
self.judge_save_fun(evol_df, file_name, self.save_root_path, fun,
raw_data_path, args)
file_name = f'continue_ud_p{n}d'
fun = 'funda_data_deal.BaseDeal.pnd_vol_continue_ud'
raw_data_path = ('EM_Funda/DERIVED_14/aadj_r.csv', )
args = (n, )
self.judge_save_fun(vol_continue_ud_df, file_name,
self.save_root_path, fun, raw_data_path, args)
def turn_pnd(self, turnratio, sector_df, n, percent):
turnratio_mean = bt.AZ_Rolling(turnratio, n).mean()
target_df = self.row_extre(turnratio_mean, sector_df, percent)
return target_df * sector_df
def wgt_return_pnd(self, aadj_r, turnratio, sector_df, n, percent):
aadj_r_c = (aadj_r * turnratio)
wgt_return_pnd_df = bt.AZ_Rolling(aadj_r_c, n).sum()
target_df = self.row_extre(wgt_return_pnd_df, sector_df, percent)
return target_df
def return_pnd(self, aadj_r, sector_df, n, percent):
return_pnd_df = bt.AZ_Rolling(aadj_r, n).sum()
target_df = self.row_extre(return_pnd_df, sector_df, percent)
return target_df
def pnd_volitality(adj_r, sector_df, n):
vol_df = bt.AZ_Rolling(adj_r, n).std() * (250**0.5)
vol_df[vol_df < 0.08] = 0.08
return vol_df
def pnd_evol(adj_r, sector_df, n):
vol_df = bt.AZ_Rolling(adj_r, n).std() * (250**0.5)
vol_df[vol_df < 0.08] = 0.08
evol_df = bt.AZ_Rolling(
vol_df, 30).apply(lambda x: 1 if x[-1] > 2 * x.mean() else 0)
return evol_df * sector_df
def pnd_volitality_count_down(adj_r, sector_df, n):
vol_df = bt.AZ_Rolling(adj_r, n).std() * (250**0.5) * sector_df
vol_df[vol_df < 0.08] = 0.08
return 1 / vol_df.replace(0, np.nan)