def _get_instrument_weights(system, an_ignored_variable, this_stage):
print(__file__ + ":" +
str(inspect.getframeinfo(inspect.currentframe())[:3][1]) +
":" + "Calculating instrument weights")
raw_instr_weights = this_stage.get_raw_instrument_weights()
instrument_list = list(raw_instr_weights.columns)
subsys_positions = [
this_stage.get_subsystem_position(instrument_code)
for instrument_code in instrument_list
]
subsys_positions = pd.concat(subsys_positions, axis=1).ffill()
subsys_positions.columns = instrument_list
instrument_weights = fix_weights_vs_pdm(raw_instr_weights,
subsys_positions)
weighting = system.config.instrument_weight_ewma_span
# smooth
instrument_weights = pd.ewma(instrument_weights, weighting)
return instrument_weights
def _get_instrument_weights(system, notUsed, this_stage):
this_stage.log.msg("Getting instrument weights")
raw_instr_weights = this_stage.get_raw_instrument_weights()
instrument_list = list(raw_instr_weights.columns)
subsys_positions = [
this_stage.get_subsystem_position(code)
for code in instrument_list
]
subsys_positions = pd.concat(subsys_positions, axis=1).ffill()
subsys_positions.columns = instrument_list
instrument_weights = fix_weights_vs_pdm(raw_instr_weights,
subsys_positions)
weighting_params = copy(system.config.instrument_weight_estimate)
# smooth
instrument_weights = pd.ewma(instrument_weights,
weighting_params['ewma_span'])
return instrument_weights
def get_instrument_weights(self):
"""
Get the time series of instrument weights, accounting for potentially missing positions, and weights that don't add up.
:returns: TxK pd.DataFrame containing weights, columns are instrument names, T covers all subsystem positions
"""
self.log.terse("Calculating instrument weights")
if self.use_estimated_instrument_weights():
raw_instr_weights = self.get_raw_estimated_instrument_weights()
else:
raw_instr_weights = self.get_raw_fixed_instrument_weights()
instrument_list = list(raw_instr_weights.columns)
subsys_positions = [
self.get_subsystem_position(instrument_code)
for instrument_code in instrument_list
]
subsys_positions = pd.concat(subsys_positions, axis=1).ffill()
subsys_positions.columns = instrument_list
instrument_weights = fix_weights_vs_pdm(raw_instr_weights,
subsys_positions)
smooth_weighting = self.parent.config.instrument_weight_ewma_span
# smooth
instrument_weights = instrument_weights.ewm(smooth_weighting).mean()
return instrument_weights
def get_instrument_weights(self):
"""
Get the time series of instrument weights, accounting for potentially missing positions, and weights that don't add up.
:returns: TxK pd.DataFrame containing weights, columns are instrument names, T covers all subsystem positions
"""
self.log.terse("Calculating instrument weights")
if self.use_estimated_instrument_weights():
raw_instr_weights = self.get_raw_estimated_instrument_weights()
else:
raw_instr_weights = self.get_raw_fixed_instrument_weights()
instrument_list = list(raw_instr_weights.columns)
subsys_positions = [
self.get_subsystem_position(instrument_code)
for instrument_code in instrument_list
]
subsys_positions = pd.concat(subsys_positions, axis=1).ffill()
subsys_positions.columns = instrument_list
instrument_weights = fix_weights_vs_pdm(raw_instr_weights,
subsys_positions)
smooth_weighting = self.parent.config.instrument_weight_ewma_span
# smooth
instrument_weights = instrument_weights.ewm(smooth_weighting).mean()
return instrument_weights
def _get_forecast_weights(system, instrument_code, this_stage ):
forecast_weights=this_stage.get_raw_forecast_weights(instrument_code)
rule_variation_list=list(forecast_weights.columns)
forecasts=[this_stage.get_capped_forecast(instrument_code, rule_variation_name) for rule_variation_name in rule_variation_list]
forecasts=pd.concat(forecasts, axis=1)
## adjust weights for missing data
forecast_weights=fix_weights_vs_pdm(forecast_weights, forecasts)
return forecast_weights
def _get_forecast_weights(system, instrument_code, this_stage):
this_stage.log.msg("Calculating forecast weights for %s" % (instrument_code),
instrument_code=instrument_code)
forecast_weights = this_stage.get_raw_forecast_weights(
instrument_code)
rule_variation_list = list(forecast_weights.columns)
forecasts = this_stage.get_all_forecasts(instrument_code, rule_variation_list)
# adjust weights for missing data
forecast_weights = fix_weights_vs_pdm(forecast_weights, forecasts)
return forecast_weights
def _get_clean_instrument_weights(
system, an_ignored_variable, this_stage):
raw_instr_weights = this_stage.get_raw_instrument_weights()
instrument_list = list(raw_instr_weights.columns)
subsys_positions = [this_stage.get_subsystem_position(instrument_code)
for instrument_code in instrument_list]
subsys_positions = pd.concat(subsys_positions, axis=1).ffill()
subsys_positions.columns = instrument_list
instrument_weights = fix_weights_vs_pdm(
raw_instr_weights, subsys_positions)
return instrument_weights
def _get_forecast_weights(system, instrument_code, this_stage):
this_stage.log.msg("Calculating forecast weights for %s" % instrument_code,
instrument_code=instrument_code)
## Get some useful stuff from the config
weighting_params=copy(system.config.forecast_weight_estimate)
forecast_weights = this_stage.get_raw_forecast_weights(
instrument_code)
rule_variation_list = list(forecast_weights.columns)
forecasts = this_stage.get_all_forecasts(instrument_code, rule_variation_list)
# adjust weights for missing data
forecast_weights = fix_weights_vs_pdm(forecast_weights, forecasts)
# smooth
forecast_weights = pd.ewma(forecast_weights, weighting_params['ewma_span'])
return forecast_weights
def _get_forecast_weights(system, instrument_code, this_stage):
print(__file__ + ":" + str(inspect.getframeinfo(inspect.currentframe())[:3][1]) + ":" +"Calculating forecast weights for %s" % (instrument_code))
forecast_weights = this_stage.get_raw_forecast_weights(
instrument_code)
rule_variation_list = list(forecast_weights.columns)
forecasts = this_stage.get_all_forecasts(instrument_code, rule_variation_list)
# adjust weights for missing data
# also aligns them together
forecast_weights = fix_weights_vs_pdm(forecast_weights, forecasts)
weighting=system.config.forecast_weight_ewma_span
# smooth
forecast_weights = pd.ewma(forecast_weights, weighting)
return forecast_weights
def _get_clean_instrument_weights(system, an_ignored_variable,
this_stage):
this_stage.log.terse("Calculating clean instrument weights")
raw_instr_weights = this_stage.get_raw_instrument_weights()
instrument_list = list(raw_instr_weights.columns)
subsys_positions = [
this_stage.get_subsystem_position(instrument_code)
for instrument_code in instrument_list
]
subsys_positions = pd.concat(subsys_positions, axis=1).ffill()
subsys_positions.columns = instrument_list
instrument_weights = fix_weights_vs_pdm(raw_instr_weights,
subsys_positions)
return instrument_weights
def _get_forecast_weights(system, instrument_code, this_stage):
this_stage.log.msg("Calculating forecast weights for %s" % (instrument_code),
instrument_code=instrument_code)
forecast_weights = this_stage.get_raw_forecast_weights(
instrument_code)
rule_variation_list = list(forecast_weights.columns)
forecasts = this_stage.get_all_forecasts(instrument_code, rule_variation_list)
# adjust weights for missing data
# also aligns them together
forecast_weights = fix_weights_vs_pdm(forecast_weights, forecasts)
weighting=system.config.forecast_weight_ewma_span
# smooth
forecast_weights = pd.ewma(forecast_weights, weighting)
return forecast_weights
def _get_instrument_weights(system, notUsed, this_stage):
this_stage.log.msg("Getting instrument weights")
raw_instr_weights = this_stage.get_raw_instrument_weights()
instrument_list = list(raw_instr_weights.columns)
subsys_positions = [this_stage.get_subsystem_position(code)
for code in instrument_list]
subsys_positions = pd.concat(subsys_positions, axis=1).ffill()
subsys_positions.columns = instrument_list
instrument_weights = fix_weights_vs_pdm(
raw_instr_weights, subsys_positions)
weighting_params=copy(system.config.instrument_weight_estimate)
# smooth
instrument_weights = pd.ewma(instrument_weights, weighting_params['ewma_span'])
return instrument_weights
def _get_forecast_weights(system, instrument_code, this_stage):
this_stage.log.msg("Calculating forecast weights for %s" %
(instrument_code),
instrument_code=instrument_code)
forecast_weights = this_stage.get_raw_forecast_weights(
instrument_code)
rule_variation_list = list(forecast_weights.columns)
forecasts = this_stage.get_all_forecasts(instrument_code,
rule_variation_list)
# adjust weights for missing data
# also aligns them together
forecast_weights = fix_weights_vs_pdm(forecast_weights, forecasts)
weighting = system.config.forecast_weight_ewma_span
# smooth
forecast_weights = pd.ewma(forecast_weights, weighting)
return forecast_weights
def _get_instrument_weights(
system, an_ignored_variable, this_stage):
this_stage.log.terse("Calculating instrument weights")
raw_instr_weights = this_stage.get_raw_instrument_weights()
instrument_list = list(raw_instr_weights.columns)
subsys_positions = [this_stage.get_subsystem_position(instrument_code)
for instrument_code in instrument_list]
subsys_positions = pd.concat(subsys_positions, axis=1).ffill()
subsys_positions.columns = instrument_list
instrument_weights = fix_weights_vs_pdm(
raw_instr_weights, subsys_positions)
weighting = system.config.instrument_weight_ewma_span
# smooth
instrument_weights = pd.ewma(instrument_weights, weighting)
return instrument_weights
def _get_instrument_weights(
system, an_ignored_variable, this_stage):
print(__file__ + ":" + str(inspect.getframeinfo(inspect.currentframe())[:3][1]) + ":" +"Calculating instrument weights")
raw_instr_weights = this_stage.get_raw_instrument_weights()
instrument_list = list(raw_instr_weights.columns)
subsys_positions = [this_stage.get_subsystem_position(instrument_code)
for instrument_code in instrument_list]
subsys_positions = pd.concat(subsys_positions, axis=1).ffill()
subsys_positions.columns = instrument_list
instrument_weights = fix_weights_vs_pdm(
raw_instr_weights, subsys_positions)
weighting=system.config.instrument_weight_ewma_span
# smooth
instrument_weights = pd.ewma(instrument_weights, weighting)
return instrument_weights
def _get_forecast_weights(system, instrument_code, this_stage):
print(__file__ + ":" +
str(inspect.getframeinfo(inspect.currentframe())[:3][1]) +
":" + "Calculating forecast weights for %s" %
(instrument_code))
forecast_weights = this_stage.get_raw_forecast_weights(
instrument_code)
rule_variation_list = list(forecast_weights.columns)
forecasts = this_stage.get_all_forecasts(instrument_code,
rule_variation_list)
# adjust weights for missing data
# also aligns them together
forecast_weights = fix_weights_vs_pdm(forecast_weights, forecasts)
weighting = system.config.forecast_weight_ewma_span
# smooth
forecast_weights = pd.ewma(forecast_weights, weighting)
return forecast_weights
def get_forecast_weights(self, instrument_code):
"""
Get the forecast weights
We forward fill all forecasts. We then adjust forecast weights so that
they are 1.0 in every period; after setting to zero when no forecast
is available.
:param instrument_code:
:type str:
:returns: TxK pd.DataFrame containing weights, columns are trading rule variation names, T covers all
KEY OUTPUT
>>> from systems.tests.testdata import get_test_object_futures_with_rules_and_capping
>>> from systems.basesystem import System
>>> (fcs, rules, rawdata, data, config)=get_test_object_futures_with_rules_and_capping()
>>> system=System([rawdata, rules, fcs, ForecastCombineFixed()], data, config)
>>>
>>> ## from config
>>> system.combForecast.get_forecast_weights("EDOLLAR").tail(2)
ewmac16 ewmac8
2015-12-10 0.5 0.5
2015-12-11 0.5 0.5
>>>
>>> config.forecast_weights=dict(EDOLLAR=dict(ewmac8=0.9, ewmac16=0.1))
>>> system2=System([rawdata, rules, fcs, ForecastCombineFixed()], data, config)
>>> system2.combForecast.get_forecast_weights("EDOLLAR").tail(2)
ewmac16 ewmac8
2015-12-10 0.1 0.9
2015-12-11 0.1 0.9
>>>
>>> del(config.forecast_weights)
>>> system3=System([rawdata, rules, fcs, ForecastCombineFixed()], data, config)
>>> system3.combForecast.get_forecast_weights("EDOLLAR").tail(2)
WARNING: No forecast weights - using equal weights of 0.5000 over all 2 trading rules in system
ewmac16 ewmac8
2015-12-10 0.5 0.5
2015-12-11 0.5 0.5
"""
self.log.msg("Calculating forecast weights for %s" % (instrument_code),
instrument_code=instrument_code)
# note these might include missing weights, eg too expensive, or absent
# from fixed weights
forecast_weights = self.get_raw_forecast_weights(instrument_code)
# we get the rule variations from forecast_weight columns, as if we've dropped
# expensive rules (when estimating) then get_trading_rules will give the wrong answer
rule_variation_list = list(forecast_weights.columns)
forecasts = self.get_all_forecasts(
instrument_code, rule_variation_list)
# adjust weights for missing data
# also aligns them together with forecasts
forecast_weights = fix_weights_vs_pdm(forecast_weights, forecasts)
weighting = self.parent.config.forecast_weight_ewma_span
# smooth
forecast_weights = forecast_weights.ewm(weighting).mean()
return forecast_weights
