wassname/catalyst
1
0
Fork 0
mirror of https://github.com/wassname/catalyst.git synced 2026-07-09 13:58:25 +08:00
Code Issues Packages Projects Releases Wiki Activity

BUG: fixed issue #111 related to positions update after restoring algo state

Browse Source
This commit is contained in:
Frederic Fortier
2018-01-06 19:08:47 -05:00
parent 13de3e69ef
commit 5d9708901d
2 changed files with 69 additions and 36 deletions
Download Patch File Download Diff File Expand all files Collapse all files
catalyst/examples/mean_reversion_simple.py
+3 -3
View File
@@ -37,8 +37,8 @@ def initialize(context):
context.base_price = None
context.current_day = None
context.RSI_OVERSOLD = 55
context.RSI_OVERBOUGHT = 65
context.RSI_OVERSOLD = 35
context.RSI_OVERBOUGHT = 50
context.CANDLE_SIZE = '5T'
context.start_time = time.time()
@@ -248,7 +248,7 @@ if __name__ == '__main__':
if live:
run_algorithm(
capital_base=0.03,
capital_base=0.025,
initialize=initialize,
handle_data=handle_data,
analyze=analyze,
catalyst/exchange/exchange_algorithm.py
+66 -33
View File
@@ -391,16 +391,20 @@ class ExchangeTradingAlgorithmLive(ExchangeTradingAlgorithmBase):
'before exiting the algorithm.')
algo_folder = get_algo_folder(self.algo_namespace)
folder = join(algo_folder, 'daily_perf')
folder = join(algo_folder, 'daily_performance')
files = [f for f in listdir(folder) if isfile(join(folder, f))]
daily_perf_list = []
for item in files:
filename = join(folder, item)
with open(filename, 'rb') as handle:
daily_perf_list.append(pickle.load(handle))
perf_period = pickle.load(handle)
perf_period_dict = perf_period.to_dict()
daily_perf_list.append(perf_period_dict)
stats = pd.DataFrame(daily_perf_list)
stats.set_index('period_close', drop=False, inplace=True)
self.analyze(stats)
@@ -460,44 +464,62 @@ class ExchangeTradingAlgorithmLive(ExchangeTradingAlgorithmBase):
return self._clock
def get_generator(self):
if self.trading_client is not None:
return self.trading_client.transform()
def _init_trading_client(self):
"""
This replaces Ziplines `_create_generator` method. The main difference
is that we are restoring performance tracker objects if available.
This allows us to stop/start algos without loosing their state.
perf = None
"""
if self.perf_tracker is None:
# Note from the Zipline dev:
# HACK: When running with the `run` method, we set perf_tracker to
# None so that it will be overwritten here.
tracker = self.perf_tracker = PerformanceTracker(
sim_params=self.sim_params,
trading_calendar=self.trading_calendar,
env=self.trading_environment,
)
# Set the dt initially to the period start by forcing it to change.
self.on_dt_changed(self.sim_params.start_session)
new_position_tracker = tracker.position_tracker
tracker.position_tracker = None
# Unpacking the perf_tracker and positions if available
perf = get_algo_object(
cum_perf = get_algo_object(
algo_name=self.algo_namespace,
key='cumulative_performance',
)
if cum_perf is not None:
tracker.cumulative_performance = cum_perf
# Ensure single common position tracker
tracker.position_tracker = cum_perf.position_tracker
today = pd.Timestamp.utcnow().floor('1D')
todays_perf = get_algo_object(
algo_name=self.algo_namespace,
key=today.strftime('%Y-%m-%d'),
rel_path='daily_performance',
)
if todays_perf is not None:
# Ensure single common position tracker
if tracker.position_tracker is not None:
todays_perf.position_tracker = tracker.position_tracker
else:
tracker.position_tracker = todays_perf.position_tracker
tracker.todays_performance = todays_perf
if tracker.position_tracker is None:
# Use a new position_tracker if not is found in the state
tracker.position_tracker = new_position_tracker
if not self.initialized:
# Calls the initialize function of the algorithm
self.initialize(*self.initialize_args, **self.initialize_kwargs)
self.initialized = True
# Call the simulation trading algorithm for side-effects:
# it creates the perf tracker
# TradingAlgorithm._create_generator(self, self.sim_params)
if perf is not None:
tracker.cumulative_performance = perf
period = self.perf_tracker.todays_performance
period.starting_cash = perf.ending_cash
period.starting_exposure = perf.ending_exposure
period.starting_value = perf.ending_value
# This does not seem to get updated correctly
period.position_tracker = perf.position_tracker
self.trading_client = ExchangeAlgorithmExecutor(
algo=self,
sim_params=self.sim_params,
@@ -507,6 +529,11 @@ class ExchangeTradingAlgorithmLive(ExchangeTradingAlgorithmBase):
restrictions=self.restrictions,
universe_func=self._calculate_universe,
)
def get_generator(self):
if self.trading_client is None:
self._init_trading_client()
return self.trading_client.transform()
def updated_portfolio(self):
@@ -677,11 +704,12 @@ class ExchangeTradingAlgorithmLive(ExchangeTradingAlgorithmBase):
self.frame_stats = list()
self.performance_needs_update = False
new_orders = self.perf_tracker.todays_performance.orders_by_id.keys()
if new_orders != self._last_orders:
orders = self.perf_tracker.todays_performance.orders_by_id.keys()
if orders != self._last_orders:
self.performance_needs_update = True
self._last_orders = copy.deepcopy(new_orders)
# Saving current orders to detect changes in the next frame
self._last_orders = copy.deepcopy(orders)
if self.performance_needs_update:
self.perf_tracker.update_performance()
@@ -698,7 +726,7 @@ class ExchangeTradingAlgorithmLive(ExchangeTradingAlgorithmBase):
self.portfolio_needs_update = False
log.info(
'got totals from exchanges, cash: {} positions: {}'.format(
'portfolio balances, cash: {}, positions: {}'.format(
cash, positions_value
)
)
@@ -710,18 +738,29 @@ class ExchangeTradingAlgorithmLive(ExchangeTradingAlgorithmBase):
# every bar no matter if the algorithm places an order or not.
self.validate_account_controls()
self._save_algo_state(data)
self.current_day = data.current_dt.floor('1D')
def _save_algo_state(self, data):
today = data.current_dt.floor('1D')
try:
self._save_stats_csv(self._process_stats(data))
except Exception as e:
log.warn('unable to calculate performance: {}'.format(e))
log.debug('saving cumulative performance object')
save_algo_object(
algo_name=self.algo_namespace,
key='cumulative_performance',
obj=self.perf_tracker.cumulative_performance,
)
self.current_day = data.current_dt.floor('1D')
log.debug('saving todays performance object')
save_algo_object(
algo_name=self.algo_namespace,
key=today.strftime('%Y-%m-%d'),
obj=self.perf_tracker.todays_performance,
rel_path='daily_performance'
)
def _process_stats(self, data):
today = data.current_dt.floor('1D')
@@ -764,12 +803,6 @@ class ExchangeTradingAlgorithmLive(ExchangeTradingAlgorithmBase):
start_dt=today,
end_dt=data.current_dt
)
save_algo_object(
algo_name=self.algo_namespace,
key=today.strftime('%Y-%m-%d'),
obj=daily_stats,
rel_path='daily_perf'
)
return recorded_cols