def get_forecast_matrix_over_code(
data_backtest,
stage_name="combForecast",
method_name="get_forecast_weights"):
instrument_codes = data_backtest.system.get_instrument_list()
trading_rules = data_backtest.system.rules.trading_rules()
trading_rule_names = list(trading_rules.keys())
datetime_cutoff = from_marker_to_datetime(data_backtest.timestamp)
value_dict = {}
for instrument_code in instrument_codes:
stage = getattr(data_backtest.system, stage_name)
method = getattr(stage, method_name)
value_row = method(instrument_code).ffill()[:datetime_cutoff].iloc[-1]
values_by_rule = [
value_row.get(
rule_name,
np.nan) for rule_name in trading_rule_names]
value_dict[instrument_code] = values_by_rule
value_df = pd.DataFrame(value_dict, index=trading_rule_names)
value_df = value_df.transpose()
return value_df
def get_list_of_values_by_instrument_for_config(data_backtest,
config_for_method):
instrument_codes = data_backtest.system.get_instrument_list()
datetime_cutoff = from_marker_to_datetime(data_backtest.timestamp)
stage = getattr(data_backtest.system, config_for_method.stage_name)
method = getattr(stage, config_for_method.method_name)
if config_for_method.global_bool:
# Same value regardless of instrument
value = method().ffill()[:datetime_cutoff].iloc[-1]
if config_for_method.col_selector is not None:
value = value[config_for_method.col_selector]
value_list = [value] * len(instrument_codes)
return value_list
if config_for_method.requires_code_bool:
# call for each code
if config_for_method.col_selector is not None:
value_list = [
method(instrument_code).ffill()[:datetime_cutoff].iloc[-1][
config_for_method.col_selector]
for instrument_code in instrument_codes
]
else:
value_list = [
method(instrument_code).ffill()[:datetime_cutoff].iloc[-1]
for instrument_code in instrument_codes
]
return value_list
if config_for_method.scalar_dict_bool:
value = method()[config_for_method.scalar_dict_entry]
value_list = [value] * len(instrument_codes)
return value_list
# get dataframe
value_row = method().ffill()[:datetime_cutoff].iloc[-1]
value_list = [
value_row.get(instrument_code, np.nan)
for instrument_code in instrument_codes
]
return value_list
def get_position_for_instrument_code_at_timestamp(data_backtest, data,
instrument_code):
diag_positions = diagPositions(data)
positions_over_time = diag_positions.get_position_df_for_strategy_and_instrument(
data_backtest.strategy_name, instrument_code)
if positions_over_time is missing_data:
return np.nan
datetime_cutoff = from_marker_to_datetime(data_backtest.timestamp)
positions_over_time_ffill = positions_over_time.ffill()
positions_before_cutoff = positions_over_time_ffill[:datetime_cutoff]
if len(positions_before_cutoff) == 0:
return np.nan
final_position = positions_before_cutoff.iloc[-1].position
return final_position
def get_forecast_matrix(
data_backtest, stage_name="combForecast", method_name="get_capped_forecast"
):
instrument_codes = data_backtest.system.get_instrument_list()
trading_rules = data_backtest.system.rules.trading_rules()
trading_rule_names = list(trading_rules.keys())
datetime_cutoff = from_marker_to_datetime(data_backtest.timestamp)
value_dict = {}
for rule_name in trading_rule_names:
value_dict[rule_name] = []
for instrument_code in instrument_codes:
stage = getattr(data_backtest.system, stage_name)
method = getattr(stage, method_name)
value = method(instrument_code, rule_name).ffill()[:datetime_cutoff][-1]
value_dict[rule_name].append(value)
value_df = pd.DataFrame(value_dict, index=instrument_codes)
return value_df
