Base

systematica.models.base

BaseStatArb

BaseStatArb()

Base Statistical Arbitrage class.

Method generated by attrs for class BaseStatArb.

Ancestors

• abc.ABC

Descendants

• systematica.api.models.arbitrage_index.ArbitrageBaseIndex
• systematica.api.models.arbitrage_index.ArbitrageClipIndex
• systematica.api.models.arbitrage_index.ArbitrageCumulativeIndex
• systematica.api.models.arbitrage_index.ArbitrageProbabilityIndex
• systematica.api.models.arbitrage_index.ArbitrageResetIndex
• systematica.api.models.arbitrage_index.ArbitrageRollingZscoreIndex
• systematica.api.models.meta_model.MetaModel
• systematica.api.models.momentum.AverageMomentumFactorCV
• systematica.api.models.momentum.RollingAverageMomentumFactor
• systematica.api.models.ou_process.RollingOUProcess
• systematica.api.models.range_breakout.RangeBreakout
• systematica.api.models.spread.RollingSpreadModel
• systematica.api.models.spread.RollingZscoreModel
• systematica.api.models.volatility.RollingVolatilityFactor
• systematica.api.models.volatility.VolatilityFactorCV
• systematica.api.models.volume_profile.RollingVolumeProfileFactor
• systematica.api.models.volume_profile.VolumeProfileFactorCV

Static methods

is_valid

is_valid(
    model_params: Dict[str, Any],
    signal_params: Dict[str, Any],
)

Validate model parameters.

This method checks if the model parameters are valid for the statistical arbitrage model. It should be implemented in subclasses and decorated with staticmethod to allow calling it without an instance. If the model does not require validation, it can be implemented as a no-op method with the @staticmethod decorator.

Examples:

To implement this method, define it in the subclass and use the @staticmethod decorator like so:

>>> @staticmethod
>>> def is_valid(model_params: tp.Kwargs, signal_params: tp.Kwargs):
...     # Implement validation logic here
...     pass

Parameters:

Name	Type	Default	Description
model_params	tp.Kwargs	--	Model parameters.
signal_params	tp.Kwargs	--	Signal parameters.

run_walk_forward

run_walk_forward(
    cls,
    **feature,
) ‑> systematica.walk_forward.base.BaseWalkForward

Analyzes model using rolling or cross-validation techniques.

Determine the appropriate wf selection class based on model attributes and use_rolling flag.

Parameters:

Name	Type	Default	Description
feature	tp.Kwargs	--	Configuration for a feature calculation.

Returns:

Type	Description
BaseReport	An instance of BaseWalkForwardRolling or BaseWalkForwardCV depending on the model and parameters.

run_analyzer

run_analyzer(
    cls,
    validate_model: bool = True,
    **feature,
) ‑> systematica.portfolio.analyzer.PortfolioAnalyzer

Executes the portfolio analyzer.

Combines scoring, signal generation, and portfolio simulation into a single pipeline for streamlined backtesting.

Parameters:

Name	Type	Default	Description
validate_model	bool	True	Validate model parameters. Defaults to True.
feature	tp.Kwargs	--	Configuration for a feature calculation.

Returns:

Type	Description
PortfolioAnalyzer	An instance of PortfolioAnalyzer containing the results of the backtest.

run_model

run_model(
    cls,
    **feature,
) ‑> pandas.core.frame.DataFrame

Run parameterized scores generation.

The method utilizes parameterization to run scores over various parameter combinations. The results can be merged and accessed as a single output object.

Parameters:

Name	Type	Default	Description
feature	tp.Kwargs	--	Configuration for a feature calculation.

Returns:

Type	Description
pd.DataFrame	Scores for each parameter combination.

run_signals

run_signals(
    cls,
    validate_model: bool = True,
    **feature,
) ‑> pandas.core.frame.DataFrame

Run parametrized signals generation.

The method utilizes parameterization to run signal generation over various parameter combinations. The results can be merged and accessed as a single output object.

Parameters:

Name	Type	Default	Description
validate_model	bool	True	Validate model parameters. Defaults to True.
feature	tp.Kwargs	--	Configuration for a feature calculation.

Returns:

Type	Description
pd.DataFrame	Signals for each parameter combination.

run_pipeline

run_pipeline(
    cls,
    validate_model: bool = True,
    to_numpy: bool = False,
    **feature,
) ‑> vectorbtpro.portfolio.base.Portfolio | collections.OrderedDict | pandas.core.frame.DataFrame | numpy.ndarray

Executes the portfolio pipeline. Method is parametrizable.

Combines scoring, signal generation, and portfolio simulation into a single pipeline for streamlined backtesting.

Parameters:

Name	Type	Default	Description
validate_model	bool	True	Validate model parameters. Defaults to True.
to_numpy	bool	False	If True, returns a np.ndarray. Default is False, returning a pd.DataFrame.
feature	tp.Kwargs	--	Configuration for a feature calculation.

Returns:

Type	Description
vbt.PF | OrderedDict | pd.DataFrame | tp.Array	Backtested portfolio or requested metrics.

run_expression

run_expression(
    cls,
    feature_expression: systematica.generic.base.FeatureExpression = None,
    runner_kwargs: Dict[str, Any] = None,
    validate_model: bool = True,
    to_numpy: bool = False,
    **feature,
) ‑> vectorbtpro.portfolio.base.Portfolio | collections.OrderedDict | pandas.core.frame.DataFrame | numpy.ndarray

Build an indicator class from an indicator expression.

Builds a new indicator class based on a Python expression string.

See vbt.IF.from_expr for more information.

Parameters:

Name	Type	Default	Description
feature_expression	FeatureExpression	to	None Instance of FeatureExpression. If None, take cls.get_expression method as reference. Defaults to None.
runner_kwargs	tp.Kwargs	--	Additional key-word arguments used in vbt.Data.run. Defaults to None.
validate_model	bool	True	Check if model parameters are valid. Defaults to True.
to_numpy	bool	False	If True, returns a np.ndarray. Default is False, returning a pd.DataFrame.
feature	tp.Kwargs	--	Configuration for a feature calculation.

Returns:

Type	Description
vbt.PF | OrderedDict | pd.DataFrame | tp.Array	Backtested portfolio or requested metrics.

Raises:

Type	Description
ValueError	If both feature_expression and cls.get_feature_expression are None.

run_optuna_study

run_optuna_study(
    objective: systematica.tuners.base.BaseOptunaObjective,
    feature: systematica.generic.base.Feature,
    search_space: Dict[str, Any],
    callbacks: Iterable = None,
) ‑> optuna.study.study.Study

Run hyperparameter optimization using optuna for a signal-based trading strategy.

• Optuna is used for hyperparameter tuning and efficiently finds the best parameters.
• The study prunes trials with poor results to accelerate the search process.

References:

• Optuna Documentation
• Pruning Trials

Parameters:

Name	Type	Default	Description
Objective	BaseOptunaObjective	--	Optuna custom objective object.
feature	Feature	--	The feature configuration.
search_space	dict	--	The hyperparameter search space.
callbacks	tp.Iterable	None	List of callback functions that are invoked at the end of each trial. Each function must accept two parameters with the following types in this order: Study and FrozenTrial.

Returns:

Type	Description
optuna.Study	The Optuna study object containing optimization results.

Methods

get_signals

get_signals(
    self,
    validate_model: bool = True,
    **feature,
) ‑> systematica.signals.base.Signals

Generate trading signals based on scores.

Parameters:

Name	Type	Default	Description
validate_model	bool	True	Validate model parameters. Defaults to True.
feature	tp.Kwargs	--	Configuration for a feature calculation.

Returns:

Type	Description
Signals	Generated signals.