class Trader(object):
entity_schema: EntityMixin = None
def __init__(self,
entity_ids: List[str] = None,
exchanges: List[str] = None,
codes: List[str] = None,
start_timestamp: Union[str, pd.Timestamp] = None,
end_timestamp: Union[str, pd.Timestamp] = None,
provider: str = None,
level: Union[str, IntervalLevel] = IntervalLevel.LEVEL_1DAY,
trader_name: str = None,
real_time: bool = False,
kdata_use_begin_time: bool = False,
draw_result: bool = True) -> None:
assert self.entity_schema is not None
self.logger = logging.getLogger(__name__)
if trader_name:
self.trader_name = trader_name
else:
self.trader_name = type(self).__name__.lower()
self.trading_signal_listeners: List[TradingListener] = []
self.selectors: List[TargetSelector] = []
self.entity_ids = entity_ids
self.exchanges = exchanges
self.codes = codes
self.provider = provider
# make sure the min level selector correspond to the provider and level
self.level = IntervalLevel(level)
self.real_time = real_time
if start_timestamp and end_timestamp:
self.start_timestamp = to_pd_timestamp(start_timestamp)
self.end_timestamp = to_pd_timestamp(end_timestamp)
else:
assert False
self.trading_dates = self.entity_schema.get_trading_dates(
start_date=self.start_timestamp, end_date=self.end_timestamp)
if real_time:
logger.info(
'real_time mode, end_timestamp should be future,you could set it big enough for running forever'
)
assert self.end_timestamp >= now_pd_timestamp()
self.kdata_use_begin_time = kdata_use_begin_time
self.draw_result = draw_result
self.account_service = SimAccountService(
entity_schema=self.entity_schema,
trader_name=self.trader_name,
timestamp=self.start_timestamp,
provider=self.provider,
level=self.level)
self.add_trading_signal_listener(self.account_service)
self.init_selectors(entity_ids=entity_ids,
entity_schema=self.entity_schema,
exchanges=self.exchanges,
codes=self.codes,
start_timestamp=self.start_timestamp,
end_timestamp=self.end_timestamp)
if self.selectors:
self.trading_level_asc = list(
set([
IntervalLevel(selector.level)
for selector in self.selectors
]))
self.trading_level_asc.sort()
self.logger.info(
f'trader level:{self.level},selectors level:{self.trading_level_asc}'
)
if self.level != self.trading_level_asc[0]:
raise Exception(
"trader level should be the min of the selectors")
self.trading_level_desc = list(self.trading_level_asc)
self.trading_level_desc.reverse()
self.targets_slot: TargetsSlot = TargetsSlot()
self.session = get_db_session('zvt', data_schema=TraderInfo)
self.on_start()
def on_start(self):
# run all the selectors
for selector in self.selectors:
# run for the history data at first
selector.run()
if self.entity_ids:
entity_ids = json.dumps(self.entity_ids)
else:
entity_ids = None
if self.exchanges:
exchanges = json.dumps(self.exchanges)
else:
exchanges = None
if self.codes:
codes = json.dumps(self.codes)
else:
codes = None
sim_account = TraderInfo(
id=self.trader_name,
entity_id=f'trader_zvt_{self.trader_name}',
timestamp=self.start_timestamp,
trader_name=self.trader_name,
entity_ids=entity_ids,
exchanges=exchanges,
codes=codes,
start_timestamp=self.start_timestamp,
end_timestamp=self.end_timestamp,
provider=self.provider,
level=self.level.value,
real_time=self.real_time,
kdata_use_begin_time=self.kdata_use_begin_time)
self.session.add(sim_account)
self.session.commit()
def init_selectors(self, entity_ids, entity_schema, exchanges, codes,
start_timestamp, end_timestamp):
"""
implement this to init selectors
"""
pass
def add_trading_signal_listener(self, listener):
if listener not in self.trading_signal_listeners:
self.trading_signal_listeners.append(listener)
def remove_trading_signal_listener(self, listener):
if listener in self.trading_signal_listeners:
self.trading_signal_listeners.remove(listener)
def handle_targets_slot(self, due_timestamp: pd.Timestamp,
happen_timestamp: pd.Timestamp):
"""
this function would be called in every cycle, you could overwrite it for your custom algorithm to select the
targets of different levels
the default implementation is selecting the targets in all levels
:param due_timestamp:
:param happen_timestamp:
"""
long_selected = None
short_selected = None
for level in self.trading_level_desc:
targets = self.targets_slot.get_targets(level=level)
if targets:
long_targets = set(targets[0])
short_targets = set(targets[1])
if not long_selected:
long_selected = long_targets
else:
long_selected = long_selected & long_targets
if not short_selected:
short_selected = short_targets
else:
short_selected = short_selected & short_targets
self.logger.debug('timestamp:{},long_selected:{}'.format(
due_timestamp, long_selected))
self.logger.debug('timestamp:{},short_selected:{}'.format(
due_timestamp, short_selected))
self.trade_the_targets(due_timestamp=due_timestamp,
happen_timestamp=happen_timestamp,
long_selected=long_selected,
short_selected=short_selected)
def get_current_account(self) -> AccountStats:
return self.account_service.account
def buy(self,
due_timestamp,
happen_timestamp,
entity_ids,
position_pct=1.0,
ignore_in_position=True):
if ignore_in_position:
account = self.get_current_account()
current_holdings = []
if account.positions:
current_holdings = [
position.entity_id for position in account.positions
if position != None and position.available_long > 0
]
entity_ids = set(entity_ids) - set(current_holdings)
if entity_ids:
position_pct = (1.0 / len(entity_ids)) * position_pct
for entity_id in entity_ids:
trading_signal = TradingSignal(
entity_id=entity_id,
due_timestamp=due_timestamp,
happen_timestamp=happen_timestamp,
trading_signal_type=TradingSignalType.open_long,
trading_level=self.level,
position_pct=position_pct)
self.send_trading_signal(trading_signal)
def sell(self,
due_timestamp,
happen_timestamp,
entity_ids,
position_pct=1.0):
# current position
account = self.get_current_account()
current_holdings = []
if account.positions:
current_holdings = [
position.entity_id for position in account.positions
if position != None and position.available_long > 0
]
shorted = set(current_holdings) & entity_ids
for entity_id in shorted:
trading_signal = TradingSignal(
entity_id=entity_id,
due_timestamp=due_timestamp,
happen_timestamp=happen_timestamp,
trading_signal_type=TradingSignalType.close_long,
trading_level=self.level,
position_pct=position_pct)
self.send_trading_signal(trading_signal)
def trade_the_targets(self,
due_timestamp,
happen_timestamp,
long_selected,
short_selected,
long_pct=1.0,
short_pct=1.0):
self.buy(due_timestamp=due_timestamp,
happen_timestamp=happen_timestamp,
entity_ids=long_selected,
position_pct=long_pct)
self.sell(due_timestamp=due_timestamp,
happen_timestamp=happen_timestamp,
entity_ids=short_selected,
position_pct=short_pct)
def send_trading_signal(self, signal: TradingSignal):
for listener in self.trading_signal_listeners:
try:
listener.on_trading_signal(signal)
except Exception as e:
self.logger.exception(e)
listener.on_trading_error(timestamp=signal.happen_timestamp,
error=e)
def on_finish(self):
# show the result
if self.draw_result:
import plotly.io as pio
pio.renderers.default = "browser"
reader = AccountStatsReader(trader_names=[self.trader_name])
df = reader.data_df
drawer = Drawer(main_data=NormalData(
df.copy()[['trader_name', 'timestamp', 'all_value']],
category_field='trader_name'))
drawer.draw_line(show=True)
def select_long_targets(self, long_targets: List[str]) -> List[str]:
if len(long_targets) > 10:
return long_targets[0:10]
return long_targets
def select_short_targets(self, short_targets: List[str]) -> List[str]:
if len(short_targets) > 10:
return short_targets[0:10]
return short_targets
def in_trading_date(self, timestamp):
return to_time_str(timestamp) in self.trading_dates
def on_time(self, timestamp):
self.logger.debug(f'current timestamp:{timestamp}')
def run(self):
# iterate timestamp of the min level,e.g,9:30,9:35,9.40...for 5min level
# timestamp represents the timestamp in kdata
for timestamp in self.entity_schema.get_interval_timestamps(
start_date=self.start_timestamp,
end_date=self.end_timestamp,
level=self.level):
if not self.in_trading_date(timestamp=timestamp):
continue
if self.real_time:
# all selector move on to handle the coming data
if self.kdata_use_begin_time:
real_end_timestamp = timestamp + pd.Timedelta(
seconds=self.level.to_second())
else:
real_end_timestamp = timestamp
seconds = (now_pd_timestamp() -
real_end_timestamp).total_seconds()
waiting_seconds = self.level.to_second() - seconds
# meaning the future kdata not ready yet,we could move on to check
if waiting_seconds > 0:
# iterate the selector from min to max which in finished timestamp kdata
for level in self.trading_level_asc:
if self.entity_schema.is_finished_kdata_timestamp(
timestamp=timestamp, level=level):
for selector in self.selectors:
if selector.level == level:
selector.move_on(timestamp,
self.kdata_use_begin_time,
timeout=waiting_seconds +
20)
# on_trading_open to setup the account
if self.level == IntervalLevel.LEVEL_1DAY or (
self.level != IntervalLevel.LEVEL_1DAY
and self.entity_schema.is_open_timestamp(timestamp)):
self.account_service.on_trading_open(timestamp)
self.on_time(timestamp=timestamp)
if self.selectors:
for level in self.trading_level_asc:
# in every cycle, all level selector do its job in its time
if self.entity_schema.is_finished_kdata_timestamp(
timestamp=timestamp, level=level):
all_long_targets = []
all_short_targets = []
for selector in self.selectors:
if selector.level == level:
long_targets = selector.get_open_long_targets(
timestamp=timestamp)
long_targets = self.select_long_targets(
long_targets)
short_targets = selector.get_open_short_targets(
timestamp=timestamp)
short_targets = self.select_short_targets(
short_targets)
all_long_targets += long_targets
all_short_targets += short_targets
if all_long_targets or all_short_targets:
self.targets_slot.input_targets(
level, all_long_targets, all_short_targets)
# the time always move on by min level step and we could check all level targets in the slot
# 1)the targets is generated for next interval
# 2)the acceptable price is next interval prices,you could buy it at current price if the time is before the timestamp(due_timestamp) when trading signal received
# 3)the suggest price the the close price for generating the signal(happen_timestamp)
due_timestamp = timestamp + pd.Timedelta(
seconds=self.level.to_second())
if level == self.level:
self.handle_targets_slot(
due_timestamp=due_timestamp,
happen_timestamp=timestamp)
# on_trading_close to calculate date account
if self.level == IntervalLevel.LEVEL_1DAY or (
self.level != IntervalLevel.LEVEL_1DAY
and self.entity_schema.is_close_timestamp(timestamp)):
self.account_service.on_trading_close(timestamp)
self.on_finish()
class Trader(object):
entity_schema: EntityMixin = None
def __init__(self,
region: Region,
entity_ids: List[str] = None,
exchanges: List[str] = None,
codes: List[str] = None,
start_timestamp: Union[str, pd.Timestamp] = None,
end_timestamp: Union[str, pd.Timestamp] = None,
provider: Provider = Provider.Default,
level: Union[str, IntervalLevel] = IntervalLevel.LEVEL_1DAY,
trader_name: str = None,
real_time: bool = False,
kdata_use_begin_time: bool = False,
draw_result: bool = True,
rich_mode: bool = True) -> None:
assert self.entity_schema is not None
self.logger = logging.getLogger(__name__)
if trader_name:
self.trader_name = trader_name
else:
self.trader_name = type(self).__name__.lower()
self.trading_signal_listeners: List[TradingListener] = []
self.selectors: List[TargetSelector] = []
self.entity_ids = entity_ids
self.exchanges = exchanges
self.codes = codes
self.region = region
self.provider = provider
# make sure the min level selector correspond to the provider and level
self.level = IntervalLevel(level)
self.real_time = real_time
if start_timestamp and end_timestamp:
self.start_timestamp = to_pd_timestamp(start_timestamp)
self.end_timestamp = to_pd_timestamp(end_timestamp)
else:
assert False
self.trading_dates = self.entity_schema.get_trading_dates(
start_date=self.start_timestamp, end_date=self.end_timestamp)
if real_time:
logger.info(
'real_time mode, end_timestamp should be future,you could set it big enough for running forever'
)
assert self.end_timestamp >= now_pd_timestamp(self.region)
self.kdata_use_begin_time = kdata_use_begin_time
self.draw_result = draw_result
self.rich_mode = rich_mode
self.account_service = SimAccountService(
entity_schema=self.entity_schema,
trader_name=self.trader_name,
timestamp=self.start_timestamp,
provider=self.provider,
level=self.level,
rich_mode=rich_mode)
self.register_trading_signal_listener(self.account_service)
self.init_selectors(entity_ids=entity_ids,
entity_schema=self.entity_schema,
exchanges=self.exchanges,
codes=self.codes,
start_timestamp=self.start_timestamp,
end_timestamp=self.end_timestamp)
if self.selectors:
self.trading_level_asc = list(
set([
IntervalLevel(selector.level)
for selector in self.selectors
]))
self.trading_level_asc.sort()
self.logger.info(
f'trader level:{self.level},selectors level:{self.trading_level_asc}'
)
if self.level != self.trading_level_asc[0]:
raise Exception(
"trader level should be the min of the selectors")
self.trading_level_desc = list(self.trading_level_asc)
self.trading_level_desc.reverse()
self.session = get_db_session('zvt', data_schema=TraderInfo)
self.level_map_long_targets = {}
self.level_map_short_targets = {}
self.trading_signals: List[TradingSignal] = []
self.on_start()
def on_start(self):
# run all the selectors
for selector in self.selectors:
# run for the history data at first
selector.run()
if self.entity_ids:
entity_ids = json.dumps(self.entity_ids)
else:
entity_ids = None
if self.exchanges:
exchanges = json.dumps(self.exchanges)
else:
exchanges = None
if self.codes:
codes = json.dumps(self.codes)
else:
codes = None
sim_account = TraderInfo(
id=self.trader_name,
entity_id=f'trader_zvt_{self.trader_name}',
timestamp=self.start_timestamp,
trader_name=self.trader_name,
entity_ids=entity_ids,
exchanges=exchanges,
codes=codes,
start_timestamp=self.start_timestamp,
end_timestamp=self.end_timestamp,
provider=self.provider,
level=self.level.value,
real_time=self.real_time,
kdata_use_begin_time=self.kdata_use_begin_time)
self.session.add(sim_account)
self.session.commit()
def init_selectors(self, entity_ids, entity_schema, exchanges, codes,
start_timestamp, end_timestamp):
"""
overwrite it to init selectors if you want to use selector/factor computing model or just write strategy in on_time
"""
pass
def register_trading_signal_listener(self, listener):
if listener not in self.trading_signal_listeners:
self.trading_signal_listeners.append(listener)
def deregister_trading_signal_listener(self, listener):
if listener in self.trading_signal_listeners:
self.trading_signal_listeners.remove(listener)
def set_long_targets_by_level(self, level: IntervalLevel,
targets: List[str]) -> None:
logger.debug(
f'level:{level},old long targets:{self.level_map_long_targets.get(level)},new long targets:{targets}'
)
self.level_map_long_targets[level] = targets
def set_short_targets_by_level(self, level: IntervalLevel,
targets: List[str]) -> None:
logger.debug(
f'level:{level},old short targets:{self.level_map_short_targets.get(level)},new short targets:{targets}'
)
self.level_map_short_targets[level] = targets
def get_long_targets_by_level(self, level: IntervalLevel) -> List[str]:
return self.level_map_long_targets.get(level)
def get_short_targets_by_level(self, level: IntervalLevel) -> List[str]:
return self.level_map_short_targets.get(level)
def select_long_targets_from_levels(self, timestamp):
"""
overwrite it to select long targets from multiple levels,the default implementation is selecting the targets in all level
:param timestamp:
"""
long_selected = None
for level in self.trading_level_desc:
long_targets = self.level_map_long_targets.get(level)
if long_targets:
long_targets = set(long_targets)
if not long_selected:
long_selected = long_targets
else:
long_selected = long_selected & long_targets
return long_selected
def select_short_targets_from_levels(self, timestamp):
"""
overwrite it to select short targets from multiple levels,the default implementation is selecting the targets in all level
:param timestamp:
"""
short_selected = None
for level in self.trading_level_desc:
short_targets = self.level_map_short_targets.get(level)
if short_targets:
short_targets = set(short_targets)
if not short_selected:
short_selected = short_targets
else:
short_selected = short_selected & short_targets
return short_selected
def get_current_account(self) -> AccountStats:
return self.account_service.account
def get_current_positions(self) -> List[Position]:
return self.get_current_account().positions
def long_position_control(self):
positions = self.get_current_positions()
position_pct = 1.0
if not positions:
position_pct = 0.2
elif len(positions) <= 10:
position_pct = 0.5
return position_pct
def short_position_control(self):
return 1.0
def buy(self,
due_timestamp,
happen_timestamp,
entity_ids,
ignore_in_position=True):
if ignore_in_position:
account = self.get_current_account()
current_holdings = []
if account.positions:
current_holdings = [
position.entity_id for position in account.positions
if position != None and position.available_long > 0
]
entity_ids = set(entity_ids) - set(current_holdings)
if entity_ids:
position_pct = self.long_position_control()
position_pct = (1.0 / len(entity_ids)) * position_pct
for entity_id in entity_ids:
trading_signal = TradingSignal(
entity_id=entity_id,
due_timestamp=due_timestamp,
happen_timestamp=happen_timestamp,
trading_signal_type=TradingSignalType.open_long,
trading_level=self.level,
position_pct=position_pct)
self.trading_signals.append(trading_signal)
def sell(self, due_timestamp, happen_timestamp, entity_ids):
# current position
account = self.get_current_account()
current_holdings = []
if account.positions:
current_holdings = [
position.entity_id for position in account.positions
if position != None and position.available_long > 0
]
shorted = set(current_holdings) & set(entity_ids)
if shorted:
position_pct = self.short_position_control()
for entity_id in shorted:
trading_signal = TradingSignal(
entity_id=entity_id,
due_timestamp=due_timestamp,
happen_timestamp=happen_timestamp,
trading_signal_type=TradingSignalType.close_long,
trading_level=self.level,
position_pct=position_pct)
self.trading_signals.append(trading_signal)
def trade_the_targets(self, due_timestamp, happen_timestamp, long_selected,
short_selected):
if short_selected:
self.sell(due_timestamp=due_timestamp,
happen_timestamp=happen_timestamp,
entity_ids=short_selected)
if long_selected:
self.buy(due_timestamp=due_timestamp,
happen_timestamp=happen_timestamp,
entity_ids=long_selected)
def on_finish(self, timestmap):
self.on_trading_finish(timestmap)
# show the result
if self.draw_result:
import plotly.io as pio
pio.renderers.default = "browser"
reader = AccountStatsReader(trader_names=[self.trader_name])
df = reader.data_df
drawer = Drawer(main_data=NormalData(
df.copy()[['trader_name', 'timestamp', 'all_value']],
category_field='trader_name'))
drawer.draw_line(show=True)
def filter_selector_long_targets(self, timestamp, selector: TargetSelector,
long_targets: List[str]) -> List[str]:
if len(long_targets) > 10:
return long_targets[0:10]
return long_targets
def filter_selector_short_targets(self, timestamp,
selector: TargetSelector,
short_targets: List[str]) -> List[str]:
if len(short_targets) > 10:
return short_targets[0:10]
return short_targets
def in_trading_date(self, timestamp):
return to_time_str(timestamp) in self.trading_dates
def on_time(self, timestamp):
self.logger.debug(f'current timestamp:{timestamp}')
def on_trading_signals(self, trading_signals: List[TradingSignal]):
for l in self.trading_signal_listeners:
l.on_trading_signals(trading_signals)
def on_trading_signal(self, trading_signal: TradingSignal):
for l in self.trading_signal_listeners:
try:
l.on_trading_signal(trading_signal)
except Exception as e:
self.logger.exception(e)
l.on_trading_error(timestamp=trading_signal.happen_timestamp,
error=e)
def on_trading_open(self, timestamp):
for l in self.trading_signal_listeners:
l.on_trading_open(timestamp)
def on_trading_close(self, timestamp):
for l in self.trading_signal_listeners:
l.on_trading_close(timestamp)
def on_trading_finish(self, timestamp):
for l in self.trading_signal_listeners:
l.on_trading_finish(timestamp)
def on_trading_error(self, timestamp, error):
for l in self.trading_signal_listeners:
l.on_trading_error(timestamp, error)
def run(self):
now = now_pd_timestamp(self.region)
# iterate timestamp of the min level,e.g,9:30,9:35,9.40...for 5min level
# timestamp represents the timestamp in kdata
for timestamp in self.entity_schema.get_interval_timestamps(
start_date=self.start_timestamp,
end_date=self.end_timestamp,
level=self.level):
if not self.in_trading_date(timestamp=timestamp):
continue
waiting_seconds = 0
if self.level == IntervalLevel.LEVEL_1DAY:
if is_same_date(timestamp, now):
while True:
self.logger.info(
f'time is:{now},just smoke for minutes')
time.sleep(60)
if now.hour >= 19:
waiting_seconds = 20
break
elif self.real_time:
# all selector move on to handle the coming data
if self.kdata_use_begin_time:
real_end_timestamp = timestamp + pd.Timedelta(
seconds=self.level.to_second())
else:
real_end_timestamp = timestamp
seconds = (now - real_end_timestamp).total_seconds()
waiting_seconds = self.level.to_second() - seconds
# meaning the future kdata not ready yet,we could move on to check
if waiting_seconds > 0:
# iterate the selector from min to max which in finished timestamp kdata
for level in self.trading_level_asc:
if self.entity_schema.is_finished_kdata_timestamp(
timestamp=timestamp, level=level):
for selector in self.selectors:
if selector.level == level:
selector.move_on(timestamp,
self.kdata_use_begin_time,
timeout=waiting_seconds + 20)
# on_trading_open to setup the account
if self.level >= IntervalLevel.LEVEL_1DAY or (
self.level != IntervalLevel.LEVEL_1DAY
and self.entity_schema.is_open_timestamp(timestamp)):
self.on_trading_open(timestamp)
self.on_time(timestamp=timestamp)
if self.selectors:
for level in self.trading_level_asc:
# in every cycle, all level selector do its job in its time
if self.entity_schema.is_finished_kdata_timestamp(
timestamp=timestamp, level=level):
all_long_targets = []
all_short_targets = []
for selector in self.selectors:
if selector.level == level:
long_targets = selector.get_open_long_targets(
timestamp=timestamp)
long_targets = self.filter_selector_long_targets(
timestamp=timestamp,
selector=selector,
long_targets=long_targets)
short_targets = selector.get_open_short_targets(
timestamp=timestamp)
short_targets = self.filter_selector_short_targets(
timestamp=timestamp,
selector=selector,
short_targets=short_targets)
if long_targets:
all_long_targets += long_targets
if short_targets:
all_short_targets += short_targets
if all_long_targets:
self.set_long_targets_by_level(
level, all_long_targets)
if all_short_targets:
self.set_short_targets_by_level(
level, all_short_targets)
# the time always move on by min level step and we could check all targets of levels
# 1)the targets is generated for next interval
# 2)the acceptable price is next interval prices,you could buy it at current price
# if the time is before the timestamp(due_timestamp) when trading signal received
# 3)the suggest price the the close price for generating the signal(happen_timestamp)
due_timestamp = timestamp + pd.Timedelta(
seconds=self.level.to_second())
if level == self.level:
long_selected = self.select_long_targets_from_levels(
timestamp)
short_selected = self.select_short_targets_from_levels(
timestamp)
self.logger.debug(
'timestamp:{},long_selected:{}'.format(
due_timestamp, long_selected))
self.logger.debug(
'timestamp:{},short_selected:{}'.format(
due_timestamp, short_selected))
self.trade_the_targets(
due_timestamp=due_timestamp,
happen_timestamp=timestamp,
long_selected=long_selected,
short_selected=short_selected)
if self.trading_signals:
self.on_trading_signals(self.trading_signals)
# clear
self.trading_signals = []
# on_trading_close to calculate date account
if self.level >= IntervalLevel.LEVEL_1DAY or (
self.level != IntervalLevel.LEVEL_1DAY
and self.entity_schema.is_close_timestamp(timestamp)):
self.on_trading_close(timestamp)
self.on_finish(timestamp)
def evaluate_size_from_timestamp(start_timestamp: pd.Timestamp,
end_timestamp: pd.Timestamp,
level: IntervalLevel,
one_day_trading_minutes,
trade_day=None):
"""
given from timestamp,level,one_day_trading_minutes,this func evaluate size of kdata to current.
it maybe a little bigger than the real size for fetching all the kdata.
:param start_timestamp:
:type start_timestamp: pd.Timestamp
:param level:
:type level: IntervalLevel
:param one_day_trading_minutes:
:type one_day_trading_minutes: int
"""
# if not end_timestamp:
# end_timestamp = now_pd_timestamp()
# else:
# end_timestamp = to_pd_timestamp(end_timestamp)
time_delta = end_timestamp - to_pd_timestamp(start_timestamp)
one_day_trading_seconds = one_day_trading_minutes * 60
if level == IntervalLevel.LEVEL_1MON:
if trade_day is not None:
try:
size = int(math.ceil(trade_day.index(start_timestamp) / 22))
size = 0 if size == 0 else size + 1
return size
except ValueError as _:
if start_timestamp < trade_day[-1]:
return int(math.ceil(len(trade_day) / 22))
# raise Exception("wrong start time:{}, error:{}".format(start_timestamp, e))
return int(math.ceil(time_delta.days / 30))
if level == IntervalLevel.LEVEL_1WEEK:
if trade_day is not None:
try:
size = int(math.ceil(trade_day.index(start_timestamp) / 5))
size = 0 if size == 0 else size + 1
return size
except ValueError as _:
if start_timestamp < trade_day[-1]:
return int(math.ceil(len(trade_day) / 5))
# raise Exception("wrong start time:{}, error:{}".format(start_timestamp, e))
return int(math.ceil(time_delta.days / 7))
if level == IntervalLevel.LEVEL_1DAY:
if trade_day is not None and len(trade_day) > 0:
try:
return trade_day.index(start_timestamp)
except ValueError as _:
if start_timestamp < trade_day[-1]:
return len(trade_day)
# raise Exception("wrong start time:{}, error:{}".format(start_timestamp, e))
return time_delta.days
if level == IntervalLevel.LEVEL_1HOUR:
if trade_day is not None:
start_date = start_timestamp.replace(hour=0, minute=0, second=0)
try:
days = trade_day.index(start_date)
time = datetime.datetime.time(start_timestamp)
size = (days) * 4 + int(math.ceil(count_hours_from_day(time)))
return size
except ValueError as _:
if start_date < trade_day[-1]:
return len(trade_day) * 4
# raise Exception("wrong start time:{}, error:{}".format(start_timestamp, e))
return int(math.ceil(time_delta.days * 4 * 2))
if level == IntervalLevel.LEVEL_30MIN:
if trade_day is not None:
start_date = start_timestamp.replace(hour=0, minute=0, second=0)
try:
days = trade_day.index(start_date)
time = datetime.datetime.time(start_timestamp)
size = (days) * 4 * 2 + int(
math.ceil(count_mins_from_day(time) / 5))
return size
except ValueError as _:
if start_date < trade_day[-1]:
return len(trade_day) * 4 * 2
# raise Exception("wrong start time:{}, error:{}".format(start_timestamp, e))
return int(math.ceil(time_delta.days * 4 * 2))
if level == IntervalLevel.LEVEL_15MIN:
if trade_day is not None:
start_date = start_timestamp.replace(hour=0, minute=0, second=0)
try:
days = trade_day.index(start_date)
time = datetime.datetime.time(start_timestamp)
size = (days) * 4 * 4 + int(
math.ceil(count_mins_from_day(time) / 5))
return size
except ValueError as _:
if start_date < trade_day[-1]:
return len(trade_day) * 4 * 4
# raise Exception("wrong start time:{}, error:{}".format(start_timestamp, e))
return int(math.ceil(time_delta.days * 4 * 4))
if level == IntervalLevel.LEVEL_5MIN:
if trade_day is not None:
start_date = start_timestamp.replace(hour=0, minute=0, second=0)
try:
days = trade_day.index(start_date)
time = datetime.datetime.time(start_timestamp)
size = (days) * 4 * 12 + int(
math.ceil(count_mins_from_day(time) / 5))
return size
except ValueError as _:
if start_date < trade_day[-1]:
return len(trade_day) * 4 * 12
# raise Exception("wrong start time:{}, error:{}".format(start_timestamp, e))
return int(math.ceil(time_delta.days * 4 * 12))
if level == IntervalLevel.LEVEL_1MIN:
if trade_day is not None:
start_date = start_timestamp.replace(hour=0, minute=0, second=0)
try:
days = trade_day.index(start_date)
time = datetime.datetime.time(start_timestamp)
size = (days) * 4 * 60 + count_mins_from_day(time)
return size
except ValueError as _:
if start_date < trade_day[-1]:
return len(trade_day) * 4 * 60
# raise Exception("wrong start time:{}, error:{}".format(start_timestamp, e))
return int(math.ceil(time_delta.days * 4 * 60))
if time_delta.days > 0:
seconds = (time_delta.days + 1) * one_day_trading_seconds
return int(math.ceil(seconds / level.to_second()))
else:
seconds = time_delta.total_seconds()
return min(int(math.ceil(seconds / level.to_second())),
one_day_trading_seconds / level.to_second())
def next_timestamp(current_timestamp: pd.Timestamp, level: IntervalLevel) -> pd.Timestamp:
current_timestamp = to_pd_timestamp(current_timestamp)
return current_timestamp + pd.Timedelta(seconds=level.to_second())
class Trader(object):
entity_schema: Type[TradableEntity] = None
def __init__(self,
entity_ids: List[str] = None,
exchanges: List[str] = None,
codes: List[str] = None,
start_timestamp: Union[str, pd.Timestamp] = None,
end_timestamp: Union[str, pd.Timestamp] = None,
provider: str = None,
level: Union[str, IntervalLevel] = IntervalLevel.LEVEL_1DAY,
trader_name: str = None,
real_time: bool = False,
kdata_use_begin_time: bool = False,
draw_result: bool = True,
rich_mode: bool = False,
adjust_type: AdjustType = None,
profit_threshold=(3, -0.3),
keep_history=False) -> None:
assert self.entity_schema is not None
assert start_timestamp is not None
assert end_timestamp is not None
self.logger = logging.getLogger(__name__)
if trader_name:
self.trader_name = trader_name
else:
self.trader_name = type(self).__name__.lower()
self.entity_ids = entity_ids
self.exchanges = exchanges
self.codes = codes
self.provider = provider
# make sure the min level selector correspond to the provider and level
self.level = IntervalLevel(level)
self.real_time = real_time
self.start_timestamp = to_pd_timestamp(start_timestamp)
self.end_timestamp = to_pd_timestamp(end_timestamp)
self.trading_dates = self.entity_schema.get_trading_dates(start_date=self.start_timestamp,
end_date=self.end_timestamp)
if real_time:
self.logger.info(
'real_time mode, end_timestamp should be future,you could set it big enough for running forever')
assert self.end_timestamp >= now_pd_timestamp()
self.kdata_use_begin_time = kdata_use_begin_time
self.draw_result = draw_result
self.rich_mode = rich_mode
self.adjust_type = AdjustType(adjust_type)
self.profit_threshold = profit_threshold
self.keep_history = keep_history
self.level_map_long_targets = {}
self.level_map_short_targets = {}
self.trading_signals: List[TradingSignal] = []
self.trading_signal_listeners: List[TradingListener] = []
self.selectors: List[TargetSelector] = []
self.account_service = SimAccountService(entity_schema=self.entity_schema,
trader_name=self.trader_name,
timestamp=self.start_timestamp,
provider=self.provider,
level=self.level,
rich_mode=self.rich_mode,
adjust_type=self.adjust_type,
keep_history=self.keep_history)
self.register_trading_signal_listener(self.account_service)
self.init_selectors(entity_ids=self.entity_ids, entity_schema=self.entity_schema, exchanges=self.exchanges,
codes=self.codes, start_timestamp=self.start_timestamp, end_timestamp=self.end_timestamp,
adjust_type=self.adjust_type)
if self.selectors:
self.trading_level_asc = list(set([IntervalLevel(selector.level) for selector in self.selectors]))
self.trading_level_asc.sort()
self.logger.info(f'trader level:{self.level},selectors level:{self.trading_level_asc}')
if self.level != self.trading_level_asc[0]:
raise Exception("trader level should be the min of the selectors")
self.trading_level_desc = list(self.trading_level_asc)
self.trading_level_desc.reverse()
# run selectors for history data at first
for selector in self.selectors:
selector.run()
self.on_start()
def on_start(self):
self.logger.info(f'trader:{self.trader_name} on_start')
def init_selectors(self, entity_ids, entity_schema, exchanges, codes, start_timestamp, end_timestamp,
adjust_type=None):
"""
overwrite it to init selectors if you want to use selector/factor computing model
:param adjust_type:
"""
pass
def update_targets_by_level(self, level: IntervalLevel, long_targets: List[str],
short_targets: List[str], ) -> None:
"""
the trading signals is generated in min level,before that,we should cache targets of all levels
:param level:
:param long_targets:
:param short_targets:
"""
self.logger.debug(
f'level:{level},old long targets:{self.level_map_long_targets.get(level)},new long targets:{long_targets}')
self.level_map_long_targets[level] = long_targets
self.logger.debug(
f'level:{level},old short targets:{self.level_map_short_targets.get(level)},new short targets:{short_targets}')
self.level_map_short_targets[level] = short_targets
def get_long_targets_by_level(self, level: IntervalLevel) -> List[str]:
return self.level_map_long_targets.get(level)
def get_short_targets_by_level(self, level: IntervalLevel) -> List[str]:
return self.level_map_short_targets.get(level)
def on_targets_selected_from_levels(self, timestamp) -> Tuple[List[str], List[str]]:
"""
this method's called in every min level cycle to select targets in all levels generated by the previous cycle
the default implementation is selecting the targets in all levels
overwrite it for your custom logic
:param timestamp: current event time
:return: long targets, short targets
"""
long_selected = None
short_selected = None
for level in self.trading_level_desc:
long_targets = self.level_map_long_targets.get(level)
# long must in all
if long_targets:
long_targets = set(long_targets)
if long_selected is None:
long_selected = long_targets
else:
long_selected = long_selected & long_targets
else:
long_selected = set()
short_targets = self.level_map_short_targets.get(level)
# short any
if short_targets:
short_targets = set(short_targets)
if short_selected is None:
short_selected = short_targets
else:
short_selected = short_selected | short_targets
return long_selected, short_selected
def get_current_account(self) -> AccountStats:
return self.account_service.get_current_account()
def get_current_positions(self) -> List[Position]:
return self.get_current_account().positions
def long_position_control(self):
positions = self.get_current_positions()
position_pct = 1.0
if not positions:
# 没有仓位，买2成
position_pct = 0.2
elif len(positions) <= 10:
# 小于10个持仓，买5成
position_pct = 0.5
# 买完
return position_pct
def short_position_control(self):
# 卖完
return 1.0
def on_profit_control(self):
if self.profit_threshold and self.get_current_positions():
positive = self.profit_threshold[0]
negative = self.profit_threshold[1]
close_long_entity_ids = []
for position in self.get_current_positions():
if position.available_long > 1:
# 止盈
if position.profit_rate >= positive:
close_long_entity_ids.append(position.entity_id)
self.logger.info(f'close profit {position.profit_rate} for {position.entity_id}')
# 止损
if position.profit_rate <= negative:
close_long_entity_ids.append(position.entity_id)
self.logger.info(f'cut lost {position.profit_rate} for {position.entity_id}')
return close_long_entity_ids, None
return None, None
def buy(self, due_timestamp, happen_timestamp, entity_ids, ignore_in_position=True):
if ignore_in_position:
account = self.get_current_account()
current_holdings = []
if account.positions:
current_holdings = [position.entity_id for position in account.positions if position != None and
position.available_long > 0]
entity_ids = set(entity_ids) - set(current_holdings)
if entity_ids:
position_pct = self.long_position_control()
position_pct = (1.0 / len(entity_ids)) * position_pct
for entity_id in entity_ids:
trading_signal = TradingSignal(entity_id=entity_id,
due_timestamp=due_timestamp,
happen_timestamp=happen_timestamp,
trading_signal_type=TradingSignalType.open_long,
trading_level=self.level,
position_pct=position_pct)
self.trading_signals.append(trading_signal)
def sell(self, due_timestamp, happen_timestamp, entity_ids):
# current position
account = self.get_current_account()
current_holdings = []
if account.positions:
current_holdings = [position.entity_id for position in account.positions if position != None and
position.available_long > 0]
shorted = set(current_holdings) & set(entity_ids)
if shorted:
position_pct = self.short_position_control()
for entity_id in shorted:
trading_signal = TradingSignal(entity_id=entity_id,
due_timestamp=due_timestamp,
happen_timestamp=happen_timestamp,
trading_signal_type=TradingSignalType.close_long,
trading_level=self.level,
position_pct=position_pct)
self.trading_signals.append(trading_signal)
def trade_the_targets(self, due_timestamp, happen_timestamp, long_selected, short_selected):
if short_selected:
self.sell(due_timestamp=due_timestamp, happen_timestamp=happen_timestamp, entity_ids=short_selected)
if long_selected:
self.buy(due_timestamp=due_timestamp, happen_timestamp=happen_timestamp, entity_ids=long_selected)
def on_finish(self, timestmap):
self.on_trading_finish(timestmap)
# show the result
if self.draw_result:
import plotly.io as pio
pio.renderers.default = "browser"
reader = AccountStatsReader(trader_names=[self.trader_name])
df = reader.data_df
drawer = Drawer(main_data=NormalData(df.copy()[['trader_name', 'timestamp', 'all_value']],
category_field='trader_name'))
drawer.draw_line(show=True)
def on_targets_filtered(self, timestamp, level, selector: TargetSelector, long_targets: List[str],
short_targets: List[str]) -> Tuple[List[str], List[str]]:
"""
overwrite it to filter the targets from selector
:param timestamp: the event time
:param level: the level
:param selector: the selector
:param long_targets: the long targets from the selector
:param short_targets: the short targets from the selector
:return: filtered long targets, filtered short targets
"""
self.logger.info(f'on_targets_filtered {level} long:{long_targets}')
if len(long_targets) > 10:
long_targets = long_targets[0:10]
self.logger.info(f'on_targets_filtered {level} filtered long:{long_targets}')
return long_targets, short_targets
def in_trading_date(self, timestamp):
return to_time_str(timestamp) in self.trading_dates
def on_time(self, timestamp: pd.Timestamp):
"""
called in every min level cycle
:param timestamp: event time
"""
self.logger.debug(f'current timestamp:{timestamp}')
def on_trading_signals(self, trading_signals: List[TradingSignal]):
for l in self.trading_signal_listeners:
l.on_trading_signals(trading_signals)
def on_trading_signal(self, trading_signal: TradingSignal):
for l in self.trading_signal_listeners:
try:
l.on_trading_signal(trading_signal)
except Exception as e:
self.logger.exception(e)
l.on_trading_error(timestamp=trading_signal.happen_timestamp, error=e)
def on_trading_open(self, timestamp):
for l in self.trading_signal_listeners:
l.on_trading_open(timestamp)
def on_trading_close(self, timestamp):
for l in self.trading_signal_listeners:
l.on_trading_close(timestamp)
def on_trading_finish(self, timestamp):
for l in self.trading_signal_listeners:
l.on_trading_finish(timestamp)
def on_trading_error(self, timestamp, error):
for l in self.trading_signal_listeners:
l.on_trading_error(timestamp, error)
def run(self):
# iterate timestamp of the min level,e.g,9:30,9:35,9.40...for 5min level
# timestamp represents the timestamp in kdata
for timestamp in self.entity_schema.get_interval_timestamps(start_date=self.start_timestamp,
end_date=self.end_timestamp, level=self.level):
if not self.in_trading_date(timestamp=timestamp):
continue
waiting_seconds = 0
if self.level == IntervalLevel.LEVEL_1DAY:
if is_same_date(timestamp, now_pd_timestamp()):
while True:
self.logger.info(f'time is:{now_pd_timestamp()},just smoke for minutes')
time.sleep(60)
current = now_pd_timestamp()
if current.hour >= 19:
waiting_seconds = 20
break
elif self.real_time:
# all selector move on to handle the coming data
if self.kdata_use_begin_time:
real_end_timestamp = timestamp + pd.Timedelta(seconds=self.level.to_second())
else:
real_end_timestamp = timestamp
seconds = (now_pd_timestamp() - real_end_timestamp).total_seconds()
waiting_seconds = self.level.to_second() - seconds
# meaning the future kdata not ready yet,we could move on to check
if waiting_seconds > 0:
# iterate the selector from min to max which in finished timestamp kdata
for level in self.trading_level_asc:
if self.entity_schema.is_finished_kdata_timestamp(timestamp=timestamp, level=level):
for selector in self.selectors:
if selector.level == level:
selector.move_on(timestamp, self.kdata_use_begin_time, timeout=waiting_seconds + 20)
# on_trading_open to setup the account
if self.level >= IntervalLevel.LEVEL_1DAY or (
self.level != IntervalLevel.LEVEL_1DAY and self.entity_schema.is_open_timestamp(timestamp)):
self.on_trading_open(timestamp)
self.on_time(timestamp=timestamp)
# 一般来说selector(factors)计算 多标的 历史数据比较快，多级别的计算也比较方便，常用于全市场标的粗过滤
# 更细节的控制可以在on_targets_filtered里进一步处理
# 也可以在on_time里面设计一些自己的逻辑配合过滤
if self.selectors:
# 多级别的遍历算法要点:
# 1)计算各级别的 标的，通过 on_targets_filtered 过滤，缓存在level_map_long_targets，level_map_short_targets
# 2)在最小的level通过 on_targets_selected_from_levels 根据多级别的缓存标的，生成最终的选中标的
# 这里需要注意的是，小级别拿到上一个周期的大级别的标的，这是合理的
for level in self.trading_level_asc:
# in every cycle, all level selector do its job in its time
if self.entity_schema.is_finished_kdata_timestamp(timestamp=timestamp, level=level):
all_long_targets = []
all_short_targets = []
# 从该level的selector中过滤targets
for selector in self.selectors:
if selector.level == level:
long_targets = selector.get_open_long_targets(timestamp=timestamp)
short_targets = selector.get_open_short_targets(timestamp=timestamp)
if long_targets or short_targets:
long_targets, short_targets = self.on_targets_filtered(timestamp=timestamp,
level=level,
selector=selector,
long_targets=long_targets,
short_targets=short_targets)
if long_targets:
all_long_targets += long_targets
if short_targets:
all_short_targets += short_targets
# 将各级别的targets缓存在level_map_long_targets，level_map_short_targets
self.update_targets_by_level(level, all_long_targets, all_short_targets)
# the time always move on by min level step and we could check all targets of levels
# 1)the targets is generated for next interval
# 2)the acceptable price is next interval prices,you could buy it at current price
# if the time is before the timestamp(due_timestamp) when trading signal received
# 3)the suggest price the the close price for generating the signal(happen_timestamp)
due_timestamp = timestamp + pd.Timedelta(seconds=self.level.to_second())
# 在最小level生成最终的 交易信号
if level == self.level:
long_selected, short_selected = self.on_targets_selected_from_levels(timestamp)
# 处理 止赢 止损
passive_short, _ = self.on_profit_control()
if passive_short:
if not short_selected:
short_selected = passive_short
else:
short_selected = list(set(short_selected) | set(passive_short))
self.logger.debug('timestamp:{},long_selected:{}'.format(due_timestamp, long_selected))
self.logger.debug('timestamp:{},short_selected:{}'.format(due_timestamp, short_selected))
if long_selected or short_selected:
self.trade_the_targets(due_timestamp=due_timestamp, happen_timestamp=timestamp,
long_selected=long_selected, short_selected=short_selected)
if self.trading_signals:
self.on_trading_signals(self.trading_signals)
# clear
self.trading_signals = []
# on_trading_close to calculate date account
if self.level >= IntervalLevel.LEVEL_1DAY or (
self.level != IntervalLevel.LEVEL_1DAY and self.entity_schema.is_close_timestamp(timestamp)):
self.on_trading_close(timestamp)
self.on_finish(timestamp)
def register_trading_signal_listener(self, listener):
if listener not in self.trading_signal_listeners:
self.trading_signal_listeners.append(listener)
def deregister_trading_signal_listener(self, listener):
if listener in self.trading_signal_listeners:
self.trading_signal_listeners.remove(listener)
