[update] 天勤tick回测

This commit is contained in:
msincenselee 2021-05-31 12:33:03 +08:00
parent a7775d5124
commit 09b57bbd7d

View File

@ -30,12 +30,22 @@ from vnpy.trader.constant import (
)
from vnpy.trader.utility import (
get_trading_date,
extract_vt_symbol,
get_underlying_symbol,
get_trading_date,
import_module_by_str
)
from .back_testing import BackTestingEngine
# vnpy交易所与淘宝数据tick目录得对应关系
VN_EXCHANGE_TICKFOLDER_MAP = {
Exchange.SHFE.value: 'SQ',
Exchange.DCE.value: 'DL',
Exchange.CZCE.value: 'ZZ',
Exchange.CFFEX.value: 'ZJ',
Exchange.INE.value: 'SQ'
}
class PortfolioTestingEngine(BackTestingEngine):
"""
@ -57,6 +67,8 @@ class PortfolioTestingEngine(BackTestingEngine):
self.bar_interval_seconds = 60 # bar csv文件属于K线类型K线的周期秒数,缺省是1分钟
self.tick_path = None # tick级别回测 路径
self.use_tq = False # True:使用tq csv数据; False:使用淘宝购买的csv数据(19年之前)
self.use_pkb2 = True # 使用tdx下载的逐笔成交数据pkb2压缩格式模拟tick
def load_bar_csv_to_df(self, vt_symbol, bar_file, data_start_date=None, data_end_date=None):
"""加载回测bar数据到DataFrame"""
@ -144,6 +156,11 @@ class PortfolioTestingEngine(BackTestingEngine):
self.output('portfolio prepare_env')
super().prepare_env(test_setting)
self.use_tq = test_setting.get('use_tq', False)
self.use_pkb2 = test_setting.get('use_pkb2', True)
if self.use_tq:
self.use_pkb2 = False
def prepare_data(self, data_dict):
"""
准备组合数据
@ -345,6 +362,138 @@ class PortfolioTestingEngine(BackTestingEngine):
traceback.print_exc()
return
def load_csv_file(self, tick_folder, vt_symbol, tick_date):
"""从文件中读取tick返回list[{dict}]"""
# 使用天勤tick数据
if self.use_tq:
return self.load_tq_csv_file(tick_folder, vt_symbol, tick_date)
# 使用淘宝下载的tick数据2019年前
symbol, exchange = extract_vt_symbol(vt_symbol)
underly_symbol = get_underlying_symbol(symbol)
exchange_folder = VN_EXCHANGE_TICKFOLDER_MAP.get(exchange.value)
if exchange == Exchange.INE:
file_path = os.path.abspath(
os.path.join(
tick_folder,
exchange_folder,
tick_date.strftime('%Y'),
tick_date.strftime('%Y%m'),
tick_date.strftime('%Y%m%d'),
'{}_{}.csv'.format(symbol.upper(), tick_date.strftime('%Y%m%d'))))
else:
file_path = os.path.abspath(
os.path.join(
tick_folder,
exchange_folder,
tick_date.strftime('%Y'),
tick_date.strftime('%Y%m'),
tick_date.strftime('%Y%m%d'),
'{}{}_{}.csv'.format(underly_symbol.upper(), symbol[-2:], tick_date.strftime('%Y%m%d'))))
ticks = []
if not os.path.isfile(file_path):
self.write_log(f'{file_path}文件不存在')
return None
df = pd.read_csv(file_path, encoding='gbk', parse_dates=False)
df.columns = ['date', 'time', 'last_price', 'volume', 'last_volume', 'open_interest',
'bid_price_1', 'bid_volume_1', 'bid_price_2', 'bid_volume_2', 'bid_price_3', 'bid_volume_3',
'ask_price_1', 'ask_volume_1', 'ask_price_2', 'ask_volume_2', 'ask_price_3', 'ask_volume_3', 'BS']
self.write_log(u'加载csv文件{}'.format(file_path))
last_time = None
for index, row in df.iterrows():
# 日期, 时间, 成交价, 成交量, 总量, 属性(持仓增减), B1价, B1量, B2价, B2量, B3价, B3量, S1价, S1量, S2价, S2量, S3价, S3量, BS
# 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
tick = row.to_dict()
tick.update({'symbol': symbol, 'exchange': exchange.value, 'trading_day': tick_date.strftime('%Y-%m-%d')})
tick_datetime = datetime.strptime(tick['date'] + ' ' + tick['time'], '%Y-%m-%d %H:%M:%S')
# 修正毫秒
if tick['time'] == last_time:
# 与上一个tick的时间去除毫秒后相同,修改为500毫秒
tick_datetime = tick_datetime.replace(microsecond=500)
tick['time'] = tick_datetime.strftime('%H:%M:%S.%f')
else:
last_time = tick['time']
tick_datetime = tick_datetime.replace(microsecond=0)
tick['time'] = tick_datetime.strftime('%H:%M:%S.%f')
tick['datetime'] = tick_datetime
# 排除涨停/跌停的数据
if (float(tick['bid_price_1']) == float('1.79769E308') and int(tick['bid_volume_1']) == 0) \
or (float(tick['ask_price_1']) == float('1.79769E308') and int(tick['ask_volume_1']) == 0):
continue
ticks.append(tick)
del df
return ticks
def load_tq_csv_file(self, tick_folder, vt_symbol, tick_date):
"""从天勤下载的csv文件中读取tick返回list[{dict}]"""
symbol, exchange = extract_vt_symbol(vt_symbol)
underly_symbol = get_underlying_symbol(symbol)
exchange_folder = VN_EXCHANGE_TICKFOLDER_MAP.get(exchange.value)
file_path = os.path.abspath(
os.path.join(
tick_folder,
tick_date.strftime('%Y%m'),
'{}_{}.csv'.format(symbol, tick_date.strftime('%Y%m%d'))))
ticks = []
if not os.path.isfile(file_path):
self.write_log(u'{}文件不存在'.format(file_path))
return None
try:
df = pd.read_csv(file_path, parse_dates=False)
# datetime,symbol,exchange,last_price,highest,lowest,volume,amount,open_interest,upper_limit,lower_limit,
# bid_price_1,bid_volume_1,ask_price_1,ask_volume_1,
# bid_price_2,bid_volume_2,ask_price_2,ask_volume_2,
# bid_price_3,bid_volume_3,ask_price_3,ask_volume_3,
# bid_price_4,bid_volume_4,ask_price_4,ask_volume_4,
# bid_price_5,bid_volume_5,ask_price_5,ask_volume_5
self.write_log(u'加载csv文件{}'.format(file_path))
last_time = None
for index, row in df.iterrows():
tick = row.to_dict()
tick['date'], tick['time'] = tick['datetime'].split(' ')
tick.update({'trading_day': tick_date.strftime('%Y-%m-%d')})
tick_datetime = datetime.strptime(tick['datetime'], '%Y-%m-%d %H:%M:%S.%f')
# 修正毫秒
if tick['time'] == last_time:
# 与上一个tick的时间去除毫秒后相同,修改为500毫秒
tick_datetime = tick_datetime.replace(microsecond=500)
tick['time'] = tick_datetime.strftime('%H:%M:%S.%f')
else:
last_time = tick['time']
tick_datetime = tick_datetime.replace(microsecond=0)
tick['time'] = tick_datetime.strftime('%H:%M:%S.%f')
tick['datetime'] = tick_datetime
# 排除涨停/跌停的数据
if (float(tick['bid_price_1']) == float('1.79769E308') and int(tick['bid_volume_1']) == 0) \
or (float(tick['ask_price_1']) == float('1.79769E308') and int(tick['ask_volume_1']) == 0):
continue
ticks.append(tick)
del df
except Exception as ex:
self.write_log(f'{file_path}文件读取不成功: {str(ex)}')
return None
return ticks
def load_bz2_cache(self, cache_folder, cache_symbol, cache_date):
"""加载缓存数据"""
if not os.path.exists(cache_folder):
@ -374,9 +523,15 @@ class PortfolioTestingEngine(BackTestingEngine):
for vt_symbol in list(self.symbol_strategy_map.keys()):
symbol, exchange = extract_vt_symbol(vt_symbol)
if self.use_pkb2:
tick_list = self.load_bz2_cache(cache_folder=self.tick_path,
cache_symbol=symbol,
cache_date=test_day.strftime('%Y%m%d'))
else:
tick_list = self.load_csv_file(tick_folder=self.tick_path,
vt_symbol=vt_symbol,
tick_date=test_day)
if not tick_list or len(tick_list) == 0:
continue
@ -417,6 +572,13 @@ class PortfolioTestingEngine(BackTestingEngine):
try:
for (dt, vt_symbol), tick_data in combined_df.iterrows():
symbol, exchange = extract_vt_symbol(vt_symbol)
last_price = tick_data.get('last_price',None)
if not last_price:
last_price = tick_data.get('price',None)
if not isinstance(last_price, float):
continue
if np.isnan(last_price):
continue
tick = TickData(
gateway_name='backtesting',
symbol=symbol,
@ -425,13 +587,10 @@ class PortfolioTestingEngine(BackTestingEngine):
date=dt.strftime('%Y-%m-%d'),
time=dt.strftime('%H:%M:%S.%f'),
trading_day=test_day.strftime('%Y-%m-%d'),
last_price=tick_data['price'],
last_price=last_price,
volume=tick_data['volume']
)
if not isinstance(tick.last_price,float):
continue
if np.isnan(tick.last_price):
continue
self.new_tick(tick)
# 结束一个交易日后,更新每日净值