def test_laststate_update_different_indexes_last_date(self):
exo_data = pd.DataFrame({
'exo': [1, 2, 3, 5, 6]
}, index=[1, 2, 3, 6, 7])
swarm_exposure = pd.Series(
[1, 2, 3, 5, 6],
index=[1, 2, 3, 5, 7]
)
swm = Swarm(self.STRATEGY_CONTEXT)
swm._last_date = 5
# Mocking
swm._swarm_series = swarm_exposure
self.assertRaises(ValueError, swm._laststate_update, exo_data, swarm_exposure)
def process(self, exo_name, swm_callback=None):
"""
Run this method every time when new EXO quote comes
:param exo_name: exo_name in MongoDB
:param swm_callback: swarm_callback after swarm is updated
:return:
"""
# Read EXO quote data from Mongo
exo_storage = EXOStorage(self.mongo_connstr, self.mongo_dbname)
# Read swarm parameters from Mongo
StrategyClass = self.strategy_context['strategy']['class']
# Load All swarms for Strategy and EXO name
swarms_data = self.load(exo_name, StrategyClass.name)
for swm_dict in swarms_data:
# Generate context for swarm
context = self.strategy_context
context['strategy']['exo_storage'] = exo_storage
context['strategy']['exo_name'] = exo_name
# Restoring swarms last state from dict
swm = Swarm.laststate_from_dict(swm_dict, context)
# Update swarm equity dynamic and last state
swm.update()
# Saving swarm state to Mongo
self.save(swm)
if swm_callback is not None:
# Run Swarm callback (to notify framework that swarm is updated)
swm_callback(swm)
def _check_context_duplicates(self, exo_name, module_name):
m = importlib.import_module('scripts.alphas.{0}.{1}'.format(
exo_name, module_name.replace('.py', '')))
existing_name = Swarm.get_name(m.STRATEGY_CONTEXT, m.STRATEGY_SUFFIX)
current_name = Swarm.get_name(self.strategy_context,
self.strategy_suffix)
if existing_name.lower() == current_name.lower():
self._log(
"Alpha strategy with similar name ('{0}') already exists, try to check code of the {1} for duplicates or change STRATEGY_SUFFIX."
.format(existing_name,
'scripts.alphas.{0}.{1}'.format(exo_name,
module_name)))
return False
self._log('Processing {0}... OK'.format(module_name))
return True
def test_laststate_update_real_high_level(self):
STRATEGY_CONTEXT = {
'strategy': {
'class': StrategyMACrossTrail,
'exo_name': './strategy_270225.mat',
'direction': -1,
'opt_params': [
# OptParam(name, default_value, min_value, max_value, step)
OptParamArray('Direction', [-1]),
OptParam('SlowMAPeriod', 20, 10, 40, 5),
OptParam('FastMAPeriod', 2, 5, 20, 5),
OptParam('MedianPeriod', 5, 2, 10, 1)
],
},
'swarm': {
'members_count': 3,
'ranking_class': RankerHighestReturns(return_period=1),
'rebalance_time_function': SwarmRebalance.every_friday
}
}
swm_start = Swarm(STRATEGY_CONTEXT)
swm_start.strategy.data = swm_start.strategy.data.ix[:'2016-03-04']
swm_start.run_swarm()
swm_start.pick()
self.assertEqual(19165.0, swm_start.picked_equity.ix['2016-03-04'])
self.assertEqual(-2, swm_start.last_exposure)
swarm_dict = swm_start.laststate_to_dict()
# Loading and updating swarm online
swm_start = Swarm.laststate_from_dict(swarm_dict, STRATEGY_CONTEXT)
swm_start.update()
self.assertEqual(19165.0, swm_start.picked_equity.ix['2016-03-04'])
self.assertAlmostEqual(19335.0, swm_start.picked_equity.ix['2016-03-07'])
self.assertAlmostEqual(18990.0, swm_start.picked_equity.ix['2016-03-10'])
self.assertAlmostEqual(20235.0, swm_start.picked_equity.ix['2016-03-11'])
def test_laststate_update_different_lengths(self):
exo_data = pd.DataFrame({
'exo': [1, 2, 3, 4, 5]
}, index=[1, 2, 3, 4, 5])
swarm_exposure = pd.Series(
[1, 2, 3, 4],
index=[1, 2, 3, 4]
)
swm = Swarm(self.STRATEGY_CONTEXT)
swm._swarm_series = None
self.assertRaises(ValueError, swm._laststate_update, exo_data, None)
swm._swarm_series = [0]
self.assertRaises(ValueError, swm._laststate_update, exo_data, None)
swm._swarm_series = [0, 2]
self.assertRaises(ValueError, swm._laststate_update, exo_data, swarm_exposure)
def init(self):
#
# Confirmation type, if confirm dialog shown
#
self.confirm_type = ''
with ui.HBox():
with ui.VBox(flex=1):
with ui.VBox():
with ui.GroupWidget(title='Deployment options'):
self.alphaname_label = ui.Label(
text=Swarm.get_name(self.strategy_context,
suffix=self.strategy_suffix), )
with ui.VBox(style='background-color: #ff8787;'
) as self.confirm_panel:
self.confirm_label = ui.Label(
text='We need your confirmation')
with ui.HBox():
self.confirm_ok = ui.Button(text='OK')
self.confirm_cancel = ui.Button(text='Cancel')
self.confirm_panel.style = 'display: none;'
self.btnrun = ui.Button(text='Deploy',
style='display: inline;')
ui.Widget(flex=1)
with ui.VBox(flex=3):
with ui.GroupWidget(flex=3,
title='Deployment progress information'):
self.log_message = ui.Label(wrap=True,
text='',
style='''overflow-y: scroll;
height: 500px;
padding: 10px;
''')
self.btnrun.visible = False
def test_laststate_update_handle_if_swarm_composition_is_empty_after_rebalance_with_costs(self, mock_get_costs):
STRATEGY_CONTEXT = {
'strategy': {
'class': StrategyMACrossTrail,
'exo_name': './mat/strategy_270225.mat',
'direction': -1,
'opt_params': [
# OptParam(name, default_value, min_value, max_value, step)
OptParamArray('Direction', [-1]),
OptParam('SlowMAPeriod', 20, 10, 40, 5),
OptParam('FastMAPeriod', 2, 5, 20, 5),
OptParam('MedianPeriod', 5, 2, 10, 1)
],
},
'swarm': {
'members_count': 3,
'ranking_class': RankerHighestReturns(return_period=1),
'rebalance_time_function': SwarmRebalance.every_friday
},
}
#
# Mocking the CostsManagerEXOFixed.get_costs
#
from backtester.matlab import loaddata
exo_df, info = loaddata('./mat/strategy_270225.mat')
mock_get_costs.return_value = pd.DataFrame({'rollover_costs': np.zeros(len(exo_df.index)),
'transaction_costs': np.ones(len(exo_df.index)) * 10},
index=exo_df.index)
swm_full = Swarm(STRATEGY_CONTEXT)
swm_full.strategy.data.at[pd.Timestamp('2016-03-04'), 'exo'] = swm_full.strategy.data.ix['2016-03-03']['exo']
swm_full.strategy.data.loc[:, 'delta'] = 1.0
swm_full.run_swarm()
swm_full.pick()
swm_start = Swarm(STRATEGY_CONTEXT)
swm_start.strategy.data.at[pd.Timestamp('2016-03-04'), 'exo'] = swm_full.strategy.data.ix['2016-03-03']['exo']
swm_start.strategy.data = swm_start.strategy.data.ix[:'2016-03-04']
swm_start.strategy.data.loc[:, 'delta'] = 1.0
swm_start.run_swarm()
swm_start.pick()
ctx = deepcopy(STRATEGY_CONTEXT)
ctx['strategy']['opt_preset'] = Swarm._parse_params(swm_start.last_members_list)
swm_next = Swarm(ctx)
swm_next.strategy.data.at[pd.Timestamp('2016-03-04'), 'exo'] = swm_full.strategy.data.ix['2016-03-03']['exo']
swm_next.strategy.data = swm_next.strategy.data.ix[:'2016-03-11']
swm_next.strategy.data.loc[:, 'delta'] = 1.0
swm_next.run_swarm()
# Make sure that old EXO price used
self.assertEqual(-2, swm_start.last_exposure)
self.assertEqual(swm_full.picked_equity.ix['2016-03-03'], swm_start.picked_equity.ix['2016-03-04'])
# After this run
with patch('warnings.warn') as mock_warn:
swm_start._laststate_update(swm_next.strategy.data, swm_next.raw_exposure.sum(axis=1), swm_next.strategy.costs)
self.assertTrue(mock_warn.called)
dt = '2016-03-04'
self.assertEqual(swm_full.series['equity'].ix[dt], swm_start.series['equity'].ix[dt])
self.assertEqual(swm_full.series['exposure'].ix[dt], swm_start.series['exposure'].ix[dt])
self.assertEqual(swm_full.series['costs'].ix[dt], swm_start.series['costs'].ix[dt])
self.assertEqual(swm_full.series['delta'].ix[dt], swm_start.series['delta'].ix[dt])
dt = '2016-03-07'
self.assertEqual(swm_full.series['equity'].ix[dt], swm_start.series['equity'].ix[dt])
self.assertEqual(swm_full.series['exposure'].ix[dt], swm_start.series['exposure'].ix[dt])
self.assertEqual(swm_full.series['costs'].ix[dt], swm_start.series['costs'].ix[dt])
self.assertEqual(swm_full.series['delta'].ix[dt], swm_start.series['delta'].ix[dt])
dt = '2016-03-08'
self.assertEqual(swm_full.series['equity'].ix[dt], swm_start.series['equity'].ix[dt])
self.assertEqual(swm_full.series['exposure'].ix[dt], swm_start.series['exposure'].ix[dt])
self.assertEqual(swm_full.series['costs'].ix[dt], swm_start.series['costs'].ix[dt])
self.assertEqual(swm_full.series['delta'].ix[dt], swm_start.series['delta'].ix[dt])
dt = '2016-03-09'
self.assertEqual(swm_full.series['equity'].ix[dt], swm_start.series['equity'].ix[dt])
self.assertEqual(swm_full.series['exposure'].ix[dt], swm_start.series['exposure'].ix[dt])
self.assertEqual(swm_full.series['costs'].ix[dt], swm_start.series['costs'].ix[dt])
self.assertEqual(swm_full.series['delta'].ix[dt], swm_start.series['delta'].ix[dt])
dt = '2016-03-10'
self.assertEqual(swm_full.series['equity'].ix[dt], swm_start.series['equity'].ix[dt])
self.assertEqual(swm_full.series['exposure'].ix[dt], swm_start.series['exposure'].ix[dt])
self.assertEqual(swm_full.series['costs'].ix[dt], swm_start.series['costs'].ix[dt])
self.assertEqual(swm_full.series['delta'].ix[dt], swm_start.series['delta'].ix[dt])
dt = '2016-03-11'
self.assertEqual(swm_full.series['equity'].ix[dt], swm_start.series['equity'].ix[dt])
self.assertEqual(swm_full.series['exposure'].ix[dt], swm_start.series['exposure'].ix[dt])
self.assertEqual(swm_full.series['costs'].ix[dt], swm_start.series['costs'].ix[dt])
self.assertEqual(swm_full.series['delta'].ix[dt], swm_start.series['delta'].ix[dt])
def test_laststate_save_load(self):
STRATEGY_CONTEXT = {
'strategy': {
'class': StrategyMACrossTrail,
'exo_name': './strategy_270225.mat',
'direction': -1,
'opt_params': [
# OptParam(name, default_value, min_value, max_value, step)
OptParamArray('Direction', [-1]),
OptParam('SlowMAPeriod', 20, 10, 40, 5),
OptParam('FastMAPeriod', 2, 5, 20, 5),
OptParam('MedianPeriod', 5, 2, 10, 1)
],
},
'swarm': {
'members_count': 3,
'ranking_class': RankerHighestReturns(return_period=1),
'rebalance_time_function': SwarmRebalance.every_friday
}
}
swm = Swarm(STRATEGY_CONTEXT)
swm.run_swarm()
swm.pick()
self.assertEqual(False, swm._islast_state)
s = swm.laststate_to_dict()
import pickle
self.assertEqual(s['direction'], -1)
self.assertEqual(s['alpha_name'], StrategyMACrossTrail.name)
self.assertEqual(s['exo_name'], swm.exo_name )
self.assertEqual(s['swarm_name'], swm.name )
self.assertEqual(s['swarm_series'], pickle.dumps(swm._swarm_series))
self.assertEqual(s['last_rebalance_date'], swm.last_rebalance_date)
self.assertEqual(s['last_members_list'], swm.last_members_list)
self.assertEqual(s['last_exoquote'], swm.last_exoquote )
self.assertEqual(s['last_prev_exposure'], swm.last_prev_exposure)
self.assertEqual(s['last_exposure'], swm.last_exposure)
self.assertEqual(s['last_date'], swm.last_date)
self.assertEqual(s['instrument'], swm.instrument)
self.assertEqual(s['exo_type'], swm.exo_type)
self.assertEqual(s['max_exposure'], swm.max_exposure)
swm = Swarm.laststate_from_dict(s, STRATEGY_CONTEXT)
self.assertEqual(True, swm._islast_state)
self.assertEqual(swm.direction[0], -1)
self.assertEqual(swm.strategy.name, StrategyMACrossTrail.name)
self.assertEqual(s['exo_name'], swm.exo_name)
self.assertEqual(s['swarm_name'], swm.name)
self.assertEqual(s['last_rebalance_date'], swm.last_rebalance_date)
self.assertEqual(s['last_members_list'], swm.last_members_list)
self.assertEqual(s['last_exoquote'], swm.last_exoquote)
self.assertEqual(s['last_prev_exposure'], swm.last_prev_exposure)
self.assertEqual(s['last_exposure'], swm.last_exposure)
self.assertEqual(s['last_date'], swm.last_date)
self.assertEqual(s['instrument'], swm.instrument)
self.assertEqual(s['exo_type'], swm.exo_type)
self.assertEqual(s['max_exposure'], swm.max_exposure)
eq = pickle.loads(s['swarm_series'])['equity']
self.assertEqual(True, np.all(eq == swm.picked_equity.values))
def test_laststate_update_real_with_costs(self, mock_get_costs):
from backtester.matlab import loaddata
STRATEGY_CONTEXT = {
'strategy': {
'class': StrategyMACrossTrail,
'exo_name': './mat/strategy_270225.mat',
'direction': -1,
'opt_params': [
# OptParam(name, default_value, min_value, max_value, step)
OptParamArray('Direction', [-1]),
OptParam('SlowMAPeriod', 20, 10, 40, 5),
OptParam('FastMAPeriod', 2, 5, 20, 5),
OptParam('MedianPeriod', 5, 2, 10, 1)
],
},
'swarm': {
'members_count': 3,
'ranking_class': RankerHighestReturns(return_period=1),
'rebalance_time_function': SwarmRebalance.every_friday
},
'costs': {
'manager': CostsManagerEXOFixed,
'context': {
'costs_options': 3.0,
'costs_futures': 3.0,
}
}
}
#
# Mocking the CostsManagerEXOFixed.get_costs
#
exo_df, info = loaddata('./mat/strategy_270225.mat')
mock_get_costs.return_value = pd.DataFrame({'rollover_costs': np.zeros(len(exo_df.index)),
'transaction_costs': np.ones(len(exo_df.index)) * 10}, index=exo_df.index)
swm_full = Swarm(STRATEGY_CONTEXT)
swm_full.strategy.data.loc[:, 'delta'] = 1.0
swm_full.run_swarm()
swm_full.pick()
swm_start = Swarm(STRATEGY_CONTEXT)
swm_start.strategy.data = swm_start.strategy.data.ix[:'2016-03-18']
swm_start.strategy.data.loc[:, 'delta'] = 1.0
swm_start.run_swarm()
swm_start.pick()
ctx = deepcopy(STRATEGY_CONTEXT)
ctx['strategy']['opt_preset'] = Swarm._parse_params(swm_start.last_members_list)
swm_next = Swarm(ctx)
swm_next.strategy.data = swm_next.strategy.data.ix[:'2016-03-25']
swm_next.strategy.data.loc[:, 'delta'] = 1.0
swm_next.run_swarm()
dt = '2016-03-18'
self.assertEqual(swm_full.series['equity'].ix[dt], swm_start.series['equity'].ix[dt])
self.assertEqual(swm_full.series['exposure'].ix[dt], swm_start.series['exposure'].ix[dt])
self.assertEqual(swm_full.series['costs'].ix[dt], swm_start.series['costs'].ix[dt])
self.assertEqual(np.isnan(swm_full.series['delta'].ix[dt]), np.isnan(swm_start.series['delta'].ix[dt]))
self.assertEqual(swm_start.last_exposure, swm_full.picked_exposure.sum(axis=1).ix['2016-03-18'])
# Updating swm_start (assuming that it was loaded from DB)
swm_start._laststate_update(swm_next.strategy.data, swm_next.raw_exposure.sum(axis=1), swm_next.strategy.costs)
dt = '2016-03-21'
self.assertEqual(swm_full.series['equity'].ix[dt], swm_start.series['equity'].ix[dt])
self.assertEqual(swm_full.series['exposure'].ix[dt], swm_start.series['exposure'].ix[dt])
self.assertEqual(swm_full.series['costs'].ix[dt], swm_start.series['costs'].ix[dt])
self.assertEqual(swm_full.series['delta'].ix[dt], swm_start.series['delta'].ix[dt])
dt = '2016-03-22'
self.assertEqual(swm_full.series['equity'].ix[dt], swm_start.series['equity'].ix[dt])
self.assertEqual(swm_full.series['exposure'].ix[dt], swm_start.series['exposure'].ix[dt])
self.assertEqual(swm_full.series['costs'].ix[dt], swm_start.series['costs'].ix[dt])
self.assertEqual(swm_full.series['delta'].ix[dt], swm_start.series['delta'].ix[dt])
dt = '2016-03-23'
self.assertEqual(swm_full.series['equity'].ix[dt], swm_start.series['equity'].ix[dt])
self.assertEqual(swm_full.series['exposure'].ix[dt], swm_start.series['exposure'].ix[dt])
self.assertEqual(swm_full.series['costs'].ix[dt], swm_start.series['costs'].ix[dt])
self.assertEqual(swm_full.series['delta'].ix[dt], swm_start.series['delta'].ix[dt])
dt = '2016-03-24'
self.assertEqual(swm_full.series['equity'].ix[dt], swm_start.series['equity'].ix[dt])
self.assertEqual(swm_full.series['exposure'].ix[dt], swm_start.series['exposure'].ix[dt])
self.assertEqual(swm_full.series['costs'].ix[dt], swm_start.series['costs'].ix[dt])
self.assertEqual(swm_full.series['delta'].ix[dt], swm_start.series['delta'].ix[dt])
dt = '2016-03-25'
self.assertEqual(swm_full.series['equity'].ix[dt], swm_start.series['equity'].ix[dt])
self.assertEqual(swm_full.series['exposure'].ix[dt], swm_start.series['exposure'].ix[dt])
self.assertEqual(swm_full.series['costs'].ix[dt], swm_start.series['costs'].ix[dt])
self.assertEqual(swm_full.series['delta'].ix[dt], swm_start.series['delta'].ix[dt])
def test_laststate_update_recalculated_exo_price(self):
STRATEGY_CONTEXT = {
'strategy': {
'class': StrategyMACrossTrail,
'exo_name': './mat/strategy_270225.mat',
'direction': -1,
'opt_params': [
# OptParam(name, default_value, min_value, max_value, step)
OptParamArray('Direction', [-1]),
OptParam('SlowMAPeriod', 20, 10, 40, 5),
OptParam('FastMAPeriod', 2, 5, 20, 5),
OptParam('MedianPeriod', 5, 2, 10, 1)
],
},
'swarm': {
'members_count': 3,
'ranking_class': RankerHighestReturns(return_period=1),
'rebalance_time_function': SwarmRebalance.every_friday
},
}
swm_full = Swarm(STRATEGY_CONTEXT)
swm_full.run_swarm()
swm_full.pick()
swm_start = Swarm(STRATEGY_CONTEXT)
swm_start.strategy.data.at[pd.Timestamp('2016-03-04'), 'exo'] += 10
swm_start.strategy.data = swm_start.strategy.data.ix[:'2016-03-04']
swm_start.run_swarm()
swm_start.pick()
ctx = deepcopy(STRATEGY_CONTEXT)
ctx['strategy']['opt_preset'] = Swarm._parse_params(swm_start.last_members_list)
swm_next = Swarm(ctx)
swm_next.strategy.data = swm_next.strategy.data.ix[:'2016-03-11']
swm_next.run_swarm()
# Make sure that old EXO price used
self.assertEqual(-2, swm_start.last_exposure)
self.assertEqual(swm_full.picked_equity.ix['2016-03-04'], swm_start.picked_equity.ix['2016-03-04'] - 10*swm_start.last_exposure)
# After this run
swm_start._laststate_update(swm_next.strategy.data, swm_next.raw_exposure.sum(axis=1))
self.assertEqual(swm_full.picked_equity.ix['2016-03-04'], swm_start.picked_equity.ix['2016-03-04'])
self.assertAlmostEqual(swm_full.picked_equity.ix['2016-03-07'], swm_start.picked_equity.ix['2016-03-07'])
self.assertAlmostEqual(swm_full.picked_equity.ix['2016-03-10'], swm_start.picked_equity.ix['2016-03-10'])
self.assertAlmostEqual(swm_full.picked_equity.ix['2016-03-11'], swm_start.picked_equity.ix['2016-03-11'])
self.assertEqual(-3, swm_start.last_exposure)
def test_laststate_update(self):
def reblance_every_5th(swarm):
return pd.Series(swarm.index % 5 == 0, index=swarm.index)
STRATEGY_CONTEXT = {
'strategy': {
'class': StrategyMACrossTrail,
'exo_name': './mat/strategy_270225.mat',
'direction': -1,
'opt_params': [
# OptParam(name, default_value, min_value, max_value, step)
OptParamArray('Direction', [-1]),
OptParam('SlowMAPeriod', 20, 10, 40, 5),
OptParam('FastMAPeriod', 2, 5, 20, 5),
OptParam('MedianPeriod', 5, 2, 10, 1)
],
},
'swarm': {
'members_count': 1,
'ranking_class': RankerHighestReturns(return_period=1),
'rebalance_time_function': reblance_every_5th
}
}
swm_index = np.array(range(11))
swm_values = np.array([
[ 0., 0.], # 0
[ 1., -1.],
[ 2., -2.],
[ 3., -3.],
[ 4., -4.],
[ 5., -5.], # 5
[ 6., -6.],
[ 7., -7.],
[ 8., -8.],
[ 9., -9.],
[ 6., -6.], # 10
])
exo_price = np.array([
0,
1,
2,
3,
4,
5, # 5
6,
7,
8,
9,
6 # 10
])
exposure_values = np.array([
[1., -1.], # 0
[1., -1.],
[1., -1.],
[1., -1.],
[1., -1.],
[1., -1.], # 5
[1., -1.],
[1., -1.],
[1., -1.],
[1., -1.],
[1., -1.], # 10
])
swm = Swarm(STRATEGY_CONTEXT)
swm._swarm = pd.DataFrame(swm_values, swm_index)
swm._swarm_inposition = pd.DataFrame(np.ones((11, 2)), swm_index)
swm._swarm_exposure = pd.DataFrame(exposure_values, swm_index, dtype=np.float)
swm.strategy.data = pd.DataFrame({'exo': pd.Series(exo_price, index=swm_index, dtype=np.float)})
swm.strategy.costs = None
swm_res = np.array([
0., #0 - ignored by default
0.,
0.,
0.,
0.,
0., #5 - first rebalance (pick system #0)
0., # Apply delayed rebalance we checked rebalance on #5 but change the position at #6
1.,
2.,
3.,
0., #10 - pick another systems (but keep prev system change to next day)
])
expected = pd.DataFrame(swm_res, index=swm_index)
swm.pick()
self.assertEqual(2, len(swm.rebalance_info))
self.assertEqual(5, swm.rebalance_info[0]['rebalance_date'])
self.assertEqual(10, swm.rebalance_info[1]['rebalance_date'])
for k, v in expected[0].items():
#print(k)
self.assertEqual(k, swm.picked_swarm.index[k])
self.assertEqual(v, swm.picked_swarm[0][k])
self.assertEqual(swm.last_date, 10)
self.assertEqual(swm.last_rebalance_date, 10)
self.assertEqual(swm.last_exposure, 1)
self.assertEqual(swm.last_members_list, [1])
self.assertEqual(True, np.all(swm.picked_equity.values == expected[0].values))
#
# DO swarm update with new quotes
#
swm_index = np.array(range(14))
exo_price = np.array([
0., # 0 - ignored by default
0.,
0.,
0.,
0.,
0., # 5 - first rebalance (pick system #0)
0., # Apply delayed rebalance we checked rebalance on #5 but change the position at #6
1.,
2.,
3.,
0., # 10 - pick another systems (but keep prev system change to next day)
10., # Should be added with last exposure
11.,
13.
])
swarm_exposure = np.array([
0., # 0 - ignored by default
0.,
0.,
0.,
0.,
0., # 5 - first rebalance (pick system #0)
0.,
0.,
0.,
0.,
0., # 10
2., # Should be added with last exposure
2.,
2.,
])
#swm = Swarm(STRATEGY_CONTEXT)
# Little hack
swm._last_exoquote = 0.0 # exo_price on #10
# Little hack
swm._last_exposure = -1
self.assertEqual(swm.last_date, 10)
self.assertEqual(swm.last_rebalance_date, 10)
self.assertEqual(swm.last_exposure, -1)
self.assertEqual(swm.last_exoquote, 0)
self.assertEqual(swm.last_members_list, [1])
exo_df = pd.DataFrame({'exo': exo_price}, index=swm_index)
swm._laststate_update(exo_df, pd.Series(swarm_exposure, index=swm_index))
self.assertEqual(swm.last_date, 13)
self.assertEqual(swm.last_rebalance_date, 10)
self.assertEqual(swm.last_exposure, 2)
self.assertEqual(swm.last_exoquote, 13)
self.assertEqual(swm.last_members_list, [1])
swm_res = np.array([
0., # 0 - ignored by default
0.,
0.,
0.,
0.,
0., # 5 - first rebalance (pick system #0)
0., # Apply delayed rebalance we checked rebalance on #5 but change the position at #6
1.,
2.,
3.,
0., # 10 - pick another systems (but keep prev system change to next day)
-10., # Apply delayed rebalance we checked rebalance on #10 but change the position at #11
-8.,
-4.,
])
expected = pd.Series(swm_res, index=swm_index)
self.assertEqual(len(swm.picked_equity), len(expected))
for k, v in expected.items():
print(k)
self.assertEqual(k, swm.picked_equity.index[k])
self.assertEqual(v, swm.picked_equity.values[k])
def test_pick_equity_calculation(self):
def reblance_every_5th(swarm):
return pd.Series(swarm.index % 5 == 0, index=swarm.index)
STRATEGY_CONTEXT = {
'strategy': {
'class': StrategyMACrossTrail,
'exo_name': './strategy_270225.mat',
'direction': -1,
'opt_params': [
# OptParam(name, default_value, min_value, max_value, step)
OptParamArray('Direction', [-1]),
OptParam('SlowMAPeriod', 20, 10, 40, 5),
OptParam('FastMAPeriod', 2, 5, 20, 5),
OptParam('MedianPeriod', 5, 2, 10, 1)
],
},
'swarm': {
'members_count': 1,
'ranking_class': RankerHighestReturns(return_period=1),
'rebalance_time_function': reblance_every_5th
}
}
swm_index = np.array(range(20))
swm_values = np.array([
[ 0., 0.], # 0
[ 1., -1.],
[ 2., -2.],
[ 3., -3.],
[ 4., -4.],
[ 5., -5.], # 5
[ 6., -6.],
[ 7., -7.],
[ 8., -8.],
[ 9., -9.],
[ 6., -6.], # 10
[ 5., -5.],
[ 4., -4.],
[ 3., -3.],
[ 2., -2.],
[ 1., -1.], #15
[ 0., 0.],
[ -1., 1.],
[ -2., 2.],
[ -3., 3.]])
exo_price = np.array([
0,
1,
2,
3,
4,
5, # 5
6,
7,
8,
9,
6, # 10
5,
4,
3,
2,
1,
0,
-1,
-2,
-3
])
exposure_values = np.array([
[1., -1.], # 0
[1., -1.],
[1., -1.],
[1., -1.],
[1., -1.],
[1., -1.], # 5
[1., -1.],
[1., -1.],
[1., -1.],
[1., -1.],
[1., -1.], # 10
[1., -1.],
[1., -1.],
[1., -1.],
[1., -1.],
[1., -1.],
[1., -1.],
[1., -1.],
[1., -1.],
[1., -1.],
])
swm = Swarm(STRATEGY_CONTEXT)
swm._swarm = pd.DataFrame(swm_values, swm_index)
swm._swarm_inposition = pd.DataFrame(np.ones((20, 2)), swm_index)
swm._swarm_exposure = pd.DataFrame(exposure_values, swm_index, dtype=np.float)
swm.strategy.data = pd.DataFrame({'exo': pd.Series(exo_price, index=swm_index, dtype=np.float)})
swm.strategy.costs = None
swm_res = np.array([
[ 0.], #0 - ignored by default
[ 0.],
[ 0.],
[ 0.],
[ 0.],
[ 0.], #5 - first rebalance (pick system #0)
[ 0.], # Apply delayed rebalance we checked rebalance on #5 but change the position at #6
[ 1.],
[ 2.],
[ 3.],
[ 0.], #10 - pick another systems (but keep prev system change to next day)
[ -1.], # Apply delayed rebalance we checked rebalance on #10 but change the position at #11
[ 0.],
[ 1.],
[ 2.],
[ 3.], #15 - keep system #1
[ 4.],
[ 5.],
[ 6.],
[ 7.]] )
expected = pd.DataFrame(swm_res, index=swm_index)
swm.pick()
for k, v in expected[0].items():
print(k)
self.assertEqual(k, swm.picked_swarm.index[k])
self.assertEqual(v, swm.picked_swarm[0][k])
def test_get_direction(self):
ctx = {
'strategy': {
'direction': -1,
'opt_params': [
OptParamArray('Direction', [-1]),
],
},
}
with patch('warnings.warn') as mock_warn:
Swarm.get_direction(ctx)
self.assertTrue(mock_warn.called)
ctx = {
'strategy': {
'opt_params': [
OptParamArray('Direction', [-1]),
],
},
}
self.assertEqual((-1, 'Short'), Swarm.get_direction(ctx))
ctx = {
'strategy': {
'opt_params': [
OptParamArray('Direction', [1]),
],
},
}
self.assertEqual((1, 'Long'), Swarm.get_direction(ctx))
ctx = {
'strategy': {
'opt_params': [
OptParamArray('Direction', [1, -1]),
],
},
}
self.assertEqual((0, 'Bidir'), Swarm.get_direction(ctx))
ctx = {
'strategy': {
'opt_params': [
OptParamArray('AnotherOptParam', [1, -1]),
],
},
}
self.assertRaises(ValueError, Swarm.get_direction, ctx)
ctx = {
'strategy': {
'opt_params': [
OptParamArray('Direction', [1, -1, 1]),
],
},
}
self.assertRaises(ValueError, Swarm.get_direction, ctx)
ctx = {
'strategy': {
'opt_params': [
OptParamArray('Direction', [2]),
],
},
}
self.assertRaises(ValueError, Swarm.get_direction, ctx)
def property_raises(self):
swm = Swarm(self.STRATEGY_CONTEXT)
swm.series
def _custom_alpha_filename(self):
alpha_name = self.strategy_context['strategy']['class'].name
direction_name = Swarm.get_direction(self.strategy_context)[1]
return 'alpha_{0}_{1}{2}'.format(
alpha_name, direction_name, self.strategy_suffix).replace(
'-', '_').replace('.', '_').lower() + '.py'
def property_raises(self):
swm = Swarm(self.STRATEGY_CONTEXT)
swm.raw_exposure
def property_raises(self):
swm = Swarm(self.STRATEGY_CONTEXT)
swm.last_delta
def property_raises(self):
swm = Swarm(self.STRATEGY_CONTEXT)
swm.last_prev_exposure
def property_raises(self):
swm = Swarm(self.STRATEGY_CONTEXT)
swm.rebalancetime
def property_raises(self):
swm = Swarm(self.STRATEGY_CONTEXT)
swm.raw_inposition
def test_rebalanceinformation_before_next_day(self):
def riseup(swarm_slice, nsystems):
result = []
rank_info = []
# Calculate 14-period equity returns and sort values
last_diff = swarm_slice.diff(periods=1).iloc[-1, :].sort_values(ascending=False)
# Pick best nsystems
best = last_diff[:nsystems]
for k, v in best.items():
if not np.isnan(v) and v > 0:
result.append(k)
rank_info.append({'rank_value': v})
return result, rank_info
def reblance_every_5th(swarm):
return pd.Series(swarm.index % 5 == 0, index=swarm.index)
STRATEGY_CONTEXT = {
'strategy': {
'class': StrategyMACrossTrail,
'exo_name': 'strategy_270225',
'direction': -1,
'opt_params': [
# OptParam(name, default_value, min_value, max_value, step)
OptParamArray('Direction', [-1]),
OptParam('SlowMAPeriod', 20, 10, 40, 5),
OptParam('FastMAPeriod', 2, 5, 20, 5),
OptParam('MedianPeriod', 5, 2, 10, 1)
],
},
'swarm': {
'members_count': 1,
'ranking_class': RankerHighestReturns(return_period=1),
'rebalance_time_function': reblance_every_5th
}
}
swm_index = np.array(range(11))
swm_values = np.array([
[ 0., 0.], # 0
[ 1., -1.],
[ 2., -2.],
[ 3., -3.],
[ 4., -4.],
[ 5., -5.], # 5
[ 6., -6.],
[ 7., -7.],
[ 8., -8.],
[ 9., -9.],
[ 6., -6.], # 10
])
exposure_values = np.array([
[1., -1.], # 0
[1., -1.],
[1., -1.],
[1., -1.],
[1., -1.],
[1., -1.], # 5
[1., -1.],
[1., -1.],
[1., -1.],
[1., -1.],
[1., -1.], # 10
])
exo_price = np.array([
0,
1,
2,
3,
4,
5, # 5
6,
7,
8,
9,
6 #10
])
swm = Swarm(STRATEGY_CONTEXT)
swm._swarm = pd.DataFrame(swm_values, swm_index)
swm._swarm_inposition = pd.DataFrame(np.ones((11, 2)), swm_index)
swm._swarm_exposure = pd.DataFrame(exposure_values, swm_index, dtype=np.float)
swm.strategy.data = pd.DataFrame({'exo': pd.Series(exo_price, index=swm_index, dtype=np.float)})
swm.strategy.costs = None
#equity1, _st_pass = stats_exposure(swm.strategy.data['exo'], swm._swarm_exposure[1], costs=None)
swm_res = np.array([
[ 0.], #0 - ignored by default
[ 0.],
[ 0.],
[ 0.],
[ 0.],
[ 0.], #5 - first rebalance (pick system #0)
[ 0.], # Apply delayed rebalance we checked rebalance on #5 but change the position at #6
[ 1.],
[ 2.],
[ 3.],
[ 0.], #10 - pick another systems (but keep prev system change to next day)
])
expected = pd.DataFrame(swm_res, index=swm_index)
swm.pick()
self.assertEqual(2, len(swm.rebalance_info))
self.assertEqual(5, swm.rebalance_info[0]['rebalance_date'])
self.assertEqual(10, swm.rebalance_info[1]['rebalance_date'])
for k, v in expected[0].items():
print(k)
self.assertEqual(k, swm.picked_swarm.index[k])
self.assertEqual(v, swm.picked_swarm[0][k])
def property_raises(self):
swm = Swarm(self.STRATEGY_CONTEXT)
swm.last_members_list
def property_raises(self):
swm = Swarm(self.STRATEGY_CONTEXT)
swm.picked_stats
def property_raises(self):
swm = Swarm(self.STRATEGY_CONTEXT)
swm.last_rebalance_date
def main(args, loglevel):
if args.logfile == '':
logging.basicConfig(
stream=sys.stdout,
format='%(asctime)s - %(name)s - %(levelname)s - %(message)s',
level=loglevel)
else:
logging.basicConfig(
filename=args.logfile,
format='%(asctime)s - %(name)s - %(levelname)s - %(message)s',
level=loglevel)
signalapp = SignalApp("AlphaRebalancer", APPCLASS_ALPHA, RABBIT_HOST,
RABBIT_USER, RABBIT_PASSW)
signalapp.send(MsgStatus('INIT', 'Initiating alpha rebalancer script'))
#exo_names = get_exo_names_mat()
logging.getLogger("pika").setLevel(logging.WARNING)
logging.info("Starting...")
exo_storage = EXOStorage(MONGO_CONNSTR, MONGO_EXO_DB)
exo_names = exo_storage.exo_list()
for exo in exo_names:
logging.info("Processing EXO: " + exo)
# Load alpha modules to process
for module in os.listdir('alphas'):
#
# Custom EXO folder found
#
swm = None
context = None
if module.lower() == exo.lower() and os.path.isdir(
os.path.join('alphas', module)):
for custom_file in os.listdir(os.path.join('alphas', module)):
if 'alpha_' in custom_file and '.py' in custom_file:
logging.debug(
'Processing custom module: ' +
os.path.join('alphas', module, custom_file))
try:
m = importlib.import_module(
'scripts.alphas.{0}.{1}'.format(
module, custom_file.replace('.py', '')))
context = m.STRATEGY_CONTEXT
context['strategy']['exo_name'] = exo
context['strategy'][
'suffix'] = m.STRATEGY_SUFFIX + 'custom'
context['strategy']['exo_storage'] = exo_storage
logging.info(
'Running CUSTOM alpha: ' + Swarm.get_name(
m.STRATEGY_CONTEXT, m.STRATEGY_SUFFIX))
if 'exo_name' in context['strategy'] and context[
'strategy']['exo_name'] != exo:
logging.error(
"Custom strategy context exo_name != current EXO name (folder mismatch?)"
)
raise ValueError(
"Custom strategy context exo_name != current EXO name (folder mismatch?)"
)
swm = Swarm(context)
swm.run_swarm()
swm.pick()
#
# Saving last EXO state to online DB
#
swmonline = SwarmOnlineManager(
MONGO_CONNSTR, MONGO_EXO_DB,
m.STRATEGY_CONTEXT)
logging.debug('Saving: {0}'.format(swm.name))
swmonline.save(swm)
except:
logging.exception('Exception occurred:')
signalapp.send(
MsgStatus('ERROR',
'Exception in {0}'.format(
Swarm.get_name(
m.STRATEGY_CONTEXT,
m.STRATEGY_SUFFIX)),
notify=True))
elif 'alpha_' in module and '.py' in module:
logging.debug('Processing generic module: ' + module)
try:
m = importlib.import_module('scripts.alphas.{0}'.format(
module.replace('.py', '')))
for direction in [-1, 1]:
context = m.STRATEGY_CONTEXT
context['strategy']['exo_name'] = exo
context['strategy']['opt_params'][0] = OptParamArray(
'Direction', [direction])
context['strategy']['suffix'] = m.STRATEGY_SUFFIX
context['strategy']['exo_storage'] = exo_storage
logging.info('Running alpha: ' +
Swarm.get_name(m.STRATEGY_CONTEXT) +
' Direction: {0}'.format(direction))
swm = Swarm(context)
swm.run_swarm()
swm.pick()
#
# Saving last EXO state to online DB
#
swmonline = SwarmOnlineManager(MONGO_CONNSTR,
MONGO_EXO_DB,
m.STRATEGY_CONTEXT)
logging.debug('Saving: {0}'.format(swm.name))
swmonline.save(swm)
except:
logging.exception('Exception occurred:')
signalapp.send(
MsgStatus('ERROR',
'Exception in {0}'.format(
Swarm.get_name(m.STRATEGY_CONTEXT,
m.STRATEGY_SUFFIX)),
notify=True))
logging.info("Processing accounts positions")
assetindex = AssetIndexMongo(MONGO_CONNSTR, MONGO_EXO_DB)
datasource = DataSourceMongo(MONGO_CONNSTR,
MONGO_EXO_DB,
assetindex,
futures_limit=10,
options_limit=10,
exostorage=exo_storage)
exmgr = ExecutionManager(MONGO_CONNSTR, datasource, dbname=MONGO_EXO_DB)
exmgr.account_positions_process(write_to_db=True)
signalapp.send(MsgStatus('DONE', 'Alpha rebalancer script', notify=True))
logging.info("Done.")