Ctrl+K

EquityFactorBasket

EquityFactorBasket#

class sigtech.framework.strategies.equity_factor_basket.EquityFactorBasket

Baseclasses: SignalStrategy

EquityFactorBasket - class implementing a basket of stocks held based on factor data.

This inherits from the SignalStrategy.

However, unlike the SignalStrategy, the default allocation function automatically normalizes the signal and a sparse threshold function is used.

In addition to the inherited inputs, following dictionary inputs are available for creating objects:

  • universe_filter: Universe time series or equity filter.

  • factor_exposure_generator: Optional factor exposure object, if empty a blank one is created.

  • factors_to_weight_function: Function mapping factors to weights.

  • factors_to_weight_function_name: Function name for db persistence of factor to weight function.

  • factors_to_weight_kwargs: Extra parameters required for the factor to weight function.

  • universe_mapping: Mapping from universe identifiers to instruments, if blank reinvestment strategies will be automatically generated.

  • reinvest_dividends: Approach to reinvesting cash dividends. Either reinvest in the awarding stock (“individual_stock”), reinvest across the portfolio using the most recent explicitly set weights (“prev_set_weights”) or to reinvest across whole portfolio keeping relative weights constant after dividend is accounted for (“post_div”) which would keep market cap exposure constant.

  • reinvestment_strategy_overrides: Dictionary of overrides to use in reinvestment strategy creation.

  • use_strategy_cache: bool flag to set if strategy service should be used for reinvestment strategies.

Example object creation:

import datetime as dtm

equity_filter = sig.EquityUniverseFilter('SPX INDEX')

efb = sig.EquityFactorBasket(universe_filter=equity_filter,
                             factor_exposure_generator=factor_exposure,
                             currency='USD',
                             start_date=dtm.date(2019, 1, 15),
                             end_date=dtm.date(2019, 2, 15),
                             rebalance_frequency='1W-FRI')
convert_long_short_weight: Optional[bool]
factor_exposure_generator: Optional[FactorExposures]
factors_to_weight_function: Optional[Callable]
factors_to_weight_function_name: Optional[str]
factors_to_weight_kwargs: Optional[dict]
reinvest_dividends: Optional[Literal['prev_set_weights', 'post_div', 'individual_stock']]
reinvestment_strategy_overrides: Optional[dict]
signal_name: Optional[str]
t0_execution: Optional[bool]
universe_filter: Union[EquityUniverseFilter, Series]
universe_mapping: Optional[dict]
use_db_for_short_version: Optional[bool]
use_strategy_cache: Optional[bool]
AVAILABLE_FACTOR_TO_WEIGHT_FUNCTIONS

alias of StandardFactorToWeightFunctions

factor_exposures_on_date(date: date, exact_date: bool = False) → DataFrame

Fetch the stock-factor loadings for the specified date.

Parameters:

  • date – Date on which to analyse the factor exposures.

  • exact_date – Whether to require the factor exposures exactlty on date or whether to return the most recent data as-of the requested date (default).

Returns:

pd.DataFrame of stock-factor exposures. Indexed by factor name, columns are stock names.

strategy_initialization(dt)

Initial decision run on the start date of the strategy.

Parameters:

dt – Reference datetime.