def get_features(security_id=u"300634.XSHE", start_date='20191202', end_date='20191206', min_unit="1", tick=False,
bc='', win_len=20):
df = DataAPI.TradeCalGet(exchangeCD=u"XSHG,XSHE", beginDate=start_date.replace('-', ''),
endDate=end_date.replace('-', ''), isOpen=u"1",
field=u"calendarDate", pandas="1")
t_dates = list(set(df['calendarDate']))
df_min = DataAPI.MktBarHistDateRangeGet(securityID=security_id, startDate=start_date.replace('-', ''),
endDate=end_date.replace('-', ''),
unit=min_unit, field=u"", pandas="1")
df_bc_min = DataAPI.MktBarHistDateRangeGet(securityID=bc, startDate=start_date.replace('-', ''),
endDate=end_date.replace('-', ''),
unit=min_unit, field=u"", pandas="1")
lst = []
for date in t_dates:
_df_min = df_min[df_min.dataDate == date]
_df_bc_min = df_bc_min[df_bc_min.dataDate == date]
_df = get_features_by_date(security_id=security_id, date=date, min_unit=min_unit, tick=tick, df_min=_df_min,
df_bc_min=_df_bc_min, win_len=win_len)
lst.append(_df)
try:
ret = pd.concat(lst, axis=0)
except Exception as ex:
logger.warn('concat the feautres for security id:{0} with error:{1}'.format(security_id, ex))
ret = None
return ret
def get_features(security_id='', start_date='', end_date='', min_unit="5"):
df = DataAPI.TradeCalGet(exchangeCD=u"XSHG,XSHE",
beginDate=start_date.replace('-', ''),
endDate=end_date.replace('-', ''),
isOpen=u"1",
field=u"calendarDate",
pandas="1")
t_dates = list(df['calendarDate'])
df_min = DataAPI.MktBarHistDateRangeGet(securityID=security_id,
startDate=start_date.replace(
'-', ''),
endDate=end_date.replace('-', ''),
unit=min_unit,
field=u"",
pandas="1")
df_min['vwap'] = df_min['totalValue'] / df_min['totalVolume']
# TODO normally it should be closePrice
df_min['ret'] = df_min[['vwap'
]].rolling(2).apply(lambda x: x[-1] / x[0] - 1)
df_min['label'] = [
1 if item >= 0.0 else 0 for item in (([0.0] + list(df_min['ret']))[1:])
]
# df_min['dealPrice'] = (df_min['openPrice'] + df_min['vwap'])/2
df_min['dealPrice'] = df_min['vwap']
return df_min
def get_transaction():
df = DataAPI.MktBarHistDateRangeGet(securityID=u"002180.XSHE",
startDate=u"20191205",
endDate=u"20191231",
unit="5",
field=u"",
pandas="1")
df_agg = df.groupby('dataDate').agg({
'totalVolume': ['max', 'min', 'mean'],
})
flatten_columns = [
'{0}_{1}'.format(item[0], item[1]) for item in df_agg.columns
]
df_agg.columns = flatten_columns
df_agg = df_agg.reset_index()
df_agg['vol_diff'] = (df_agg['totalVolume_max'] - df_agg['totalVolume_min']
) / df_agg['totalVolume_mean']
df_agg.sort_values(by='vol_diff', ascending=False, inplace=True)
print(df_agg)
def use_case_sim(security_id='002180.XSHE',
participant_ratio=0.2,
start_date='20191101',
end_date='20191231',
trade_date='2019-12-05',
passive_ratio=0.5,
mode=1):
'''
mode = 1: all execute by passive(twap/vwap); mode = 2: all execute by liquididy; mode=3: ratio of passive execute is
passive_ratio, others by liquidity
'''
prev_trade_date = '2019-12-04'
df = DataAPI.MktEqudGet(
secID=security_id,
tradeDate=u"",
beginDate=start_date,
endDate=end_date,
isOpen="",
field=["tradeDate", "chgPct", "closePrice", "turnoverVol"],
pandas="1")
df['ret_std'] = df[['chgPct']].rolling(20).apply(lambda x: x.std())
df['vol_sum'] = df[['turnoverVol']].rolling(5).apply(lambda x: x.sum())
trade_row = df[df.tradeDate == trade_date]
prev_row = df[df.tradeDate == prev_trade_date]
daily_vol = list(trade_row['ret_std'])[0]
total_vol = list(prev_row['vol_sum'])[0]
passive_ratio = 1.0 if mode == 1 else (0.0 if mode == 2 else passive_ratio)
passive_vol = total_vol * passive_ratio
liquid_vol = total_vol - passive_vol
df = DataAPI.MktBarHistDateRangeGet(securityID=security_id,
startDate=trade_date.replace('-', ''),
endDate=trade_date.replace('-', ''),
unit="1",
field=u"",
pandas="1")
# df.sort_values(by='totalVolume', ascending=True, inplace=True)
df['vwap'] = df['totalValue'] / df['totalVolume']
s = passive_vol * participant_ratio / 240
adv = total_vol
sigma = daily_vol * math.sqrt(245)
df_tick = get_features_by_date(security_id=u"002180.XSHE",
date='20191205',
min_unit="1",
tick=True)
ask_vol_sum = {}
bid_vol_sum = {}
df_tick = df_tick[df_tick.barTime <= "14:57"]
df_tick = df_tick[df_tick.barTime > '09:30']
for row in (df_tick[['barTime', 'totalAskVolume',
'totalBidVolume']].values):
_ask_tmp = ask_vol_sum.get(row[0]) or 0.0
_bid_tmp = bid_vol_sum.get(row[0]) or 0.0
ask_vol_sum.update({row[0]: _ask_tmp + row[1]})
bid_vol_sum.update({row[0]: _bid_tmp + row[2]})
q_dict = {}
for k, v in bid_vol_sum.items():
q_dict.update({k: abs(v - ask_vol_sum.get(k))})
df = df[df.barTime <= "14:57"]
df = df[df.barTime > '09:30']
mkt_vols = list(df['totalVolume'])
_tmp = [s / item for item in mkt_vols]
participant_min = dict(zip(df['barTime'], _tmp))
vt_dict = dict(zip(df['barTime'], df['totalVolume']))
total_value = 0.0
total_vol = 0
# TODO get the TWAP/VWAP simulate results (with and without market impact), check market impact results
if passive_ratio > 0:
istar, tmp_dict, perm_dict = get_market_impact(s, adv, sigma, vt_dict,
q_dict)
ret_rows = []
for bar_time, item in df[['barTime', 'vwap']].values:
total_vol += s
# _tmp_impact = tmp_dict.get(bar_time)
_tmp_impact = istar
total_value += item * s * (1 - _tmp_impact / 10000)
ret_rows.append([
bar_time, istar,
tmp_dict.get(bar_time),
perm_dict.get(bar_time)
])
mi_df = pd.DataFrame(ret_rows,
columns=['barTime', 'istar', 'tmp', 'perm'])
mi_df.to_csv('mi_report_mode_{0}_{1}.csv'.format(mode, passive_ratio))
if passive_ratio < 1.0:
print('liquid target_vol is: ', liquid_vol * participant_ratio)
liquid_trade_value, liquid_trade_vol, finish_time = get_liquid_results(
df_tick,
liquid_vol * participant_ratio,
ratio=0.13,
level=1,
price_rule=1)
print(liquid_trade_value, liquid_trade_vol,
liquid_trade_value / liquid_trade_vol, finish_time)
total_value += liquid_trade_value
total_vol += liquid_trade_vol
# liquid_value, liquid_vol, finish_time = get_liquid_results(df_tick, liquid_vol * participant_ratio, ratio=0.20,
# level=2, price_rule=1)
# print(liquid_value, liquid_value, liquid_value / liquid_vol, finish_time)
print(total_value, total_vol, total_value / total_vol)
