def calc_hmo_express(code, date):
df = lut.load_orderbook(code=code, date=date, verbose=10)
if df.shape[0] > 0:
df = df[(df['timeint'] > 93000) & (df['timeint'] < 145700)].copy()
dt, co, hmo = date, code, calc_hmo(df)
else:
dt, co, hmo = date, code, np.nan
del df
gc.collect()
return date, code, hmo
def trendstrength_express(code, date):
df = lut.load_orderbook(code=code, date=date, verbose=0)
is_abnormal = np.all(df['ap1'] == df['bp1']) | np.all(
df['ap1'] == 0) | np.all(df['bp1'] == 0)
if (df.shape[0] > 0) and (~is_abnormal):
df = df[(df['timeint'] > 93000) & (df['timeint'] < 145700)].copy()
dt, co, fatv = date, code, trendstrength(df)
else:
dt, co, fatv = date, code, np.nan
del df
gc.collect()
return dt, co, fatv
def calc_dbook_express(code, date):
# TODO to be tested
df = lut.load_orderbook(code=code, date=date, verbose=0)
is_abnormal = np.all(df['ap1'] == df['bp1']) | np.all(
df['ap1'] == 0) | np.all(df['bp1'] == 0)
if (df.shape[0] > 0) and (~is_abnormal):
df = df[(df['timeint'] > 93000) & (df['timeint'] < 145700)].copy()
dt, co, fatv = date, code, dbook(df, n=1, alpha=0.05)[-1]
else:
dt, co, fatv = date, code, np.nan
del df
gc.collect()
return dt, co, fatv
test_slice = np.s_[-252:]
all_trading_dates = all_trading_dates[test_slice]
all_trading_dates_int = all_trading_dates_int[test_slice]
selected_codes1 = hs300b[0].tolist()
selected_codes2 = hs300b[2].tolist()
selected_codes3 = hs300b[3].tolist()
# --------------------------------------------------------------------------------
# compute factors and target
# --------------------------------------------------------------------------------
for date in tqdm(all_trading_dates_int):
for code in selected_codes3:
print(lfp.volume_distance_express(code, date))
df = lut.load_orderbook(code, date)
df = df[(df['timeint'] > 93000) & (df['timeint'] < 145700)].copy()
lfp.price_volume_corr_express(code, date)
raise Exception()
print('computing tba...')
tba_list = Parallel(n_jobs=16, verbose=3)(
delayed(lfp.calc_tba_express)(code=co, date=d)
for d in all_trading_dates_int for co in selected_codes3)
tba = pd.DataFrame(tba_list, columns=['date', 'code', 'tba'])
# tba['code'] = tba['code'].map(code_type_map2)
tba = tba.set_index(['date', 'code'])
print('computing dbook...')
dbook_list = Parallel(n_jobs=16, verbose=3)(
delayed(lfp.calc_dbook_express)(code=co, date=d)