Volatility

systematica.api.models.volatility

Volatility factors analyze price volatility patterns across multiple time horizons to generate trading signals based on volatility regime changes and trends.

Volatility Model Components

Stage	Main Function	Subcomponents/Components
Dynamic Window	get_dynamic_window_nb()	short_term, medium_term, long_term
Volatility Computation	get_volatilities_nb()	short_vol, medium_vol, long_vol
Score Components	scores_nb()	medium_term_impact, long_term_impact, trend_impact, direction
Final Scoring	weighted_average_scores_nb()	volatility_nb() (with tanh() amplification)

Volatility API Classes

Class	Description	Key Features
RollingVolatilityFactor	Rolling window volatility analysis	Dynamic window sizing, trend detection
VolatilityFactorCV	Cross-validated volatility analysis	Walk-forward validation, parameter stability

RollingVolatilityFactor

RollingVolatilityFactor(
    window: int = 365,
    minp: int = None,
    vol_window: int = 30,
    min_periods: int = 10,
    medium_term_weight: float = 0.5,
    long_term_weight: float = 0.2,
    trend_weight: float = 0.3,
    signal_impact_multiplier: float = 5.0,
)

Rolling Volatility variation factor.

• Mutliple Dynamic Window: Model is determines through three different time horizons — short, medium and long-term horizons.
• Volatility Insigths: The standard deviations (volatilities) over the defined dynamic windows.
• Scoring Mechanism: Relative scores between periods, trend and direction are encapsulated.
• Signal Calculation: Aggregates signals and applies a hyperbolic tangent (tanh)

Workflow:

1. Calculate Returns: compute log returns from closing prices.

2. Compute Volatility on Different Time Horizons: Derive signals through weighted average standard deviation calculated on multiple periods

3. Aggregate Signals: Combine the individual signals using their mean and compute the hyperbolic tangent element-wise.

4. Handling Multiple Symbols: The process is repeated for each pair, offering a broader market view.

Method generated by attrs for class RollingVolatilityFactor.

Ancestors

• systematica.models.base.BaseStatArb
• abc.ABC

Instance variables

• long_term_weight: float: Long term weight coeff. Defaults to 0.2.

• medium_term_weight: float: Medium term weight coeff. Defaults to 0.5.

• min_periods: int: Mininim period for calculating volatility. Defaults to 10.

• minp: int: Mininim period. Defaults to None, which means no minimum period is applied.

• signal_impact_multiplier: float: Signal impact multiplier. Defaults to 5.0.

• trend_weight: float: Trend weights. Defaults to 0.3.

• vol_window: int: Volatility window size. Defaults to 30.

• window: int: The size of the rolling window. Defaults to 365.

VolatilityFactorCV

VolatilityFactorCV(
    training_window: int = 365,
    testing_window: int = 60,
    splitter: str = 'from_custom_rolling',
    custom_splitter: str | None = None,
    custom_splitter_kwargs: dict | None = None,
    vol_window: int = 30,
    min_periods: int = 10,
    medium_term_weight: float = 0.5,
    long_term_weight: float = 0.2,
    trend_weight: float = 0.3,
    signal_impact_multiplier: float = 5.0,
)

Volatility variation factor cross validation.

• Mutliple Dynamic Window: Model is determines through three different time horizons — short, medium and long-term horizons.
• Volatility Insigths: The standard deviations (volatilities) over the defined dynamic windows.
• Scoring Mechanism: Relative scores between periods, trend and direction are encapsulated.
• Signal Calculation: Aggregates signals and applies a hyperbolic tangent (tanh)

Workflow:

1. Calculate Returns: compute log returns from closing prices.

2. Compute Volatility on Different Time Horizons: Derive signals through weighted average standard deviation calculated on multiple periods

3. Aggregate Signals: Combine the individual signals using their mean and compute the hyperbolic tangent element-wise.

4. Handling Multiple Symbols: The process is repeated for each pair, offering a broader market view.

Method generated by attrs for class VolatilityFactorCV.

Ancestors

• systematica.models.base.BaseStatArb
• abc.ABC

Instance variables

• custom_splitter: str | None: Custom splitter to use for data partitioning. Defaults to None, which means no custom splitter is used.

• custom_splitter_kwargs: dict | None: Additional keyword arguments for the custom splitter. Defaults to None, which means no additional arguments are passed.

• long_term_weight: float: Long term weight coefficient. Defaults to 0.2. This is the weight applied to the long-term volatility in the final score calculation.

• medium_term_weight: float: Medium term weight coefficient. Defaults to 0.5. This is the weight applied to the medium-term volatility in the final score calculation.

• min_periods: int: Mininim period for calculating volatility. Defaults to 10. This is the minimum number of periods required to calculate volatility.

• signal_impact_multiplier: float: Signal impact multiplier. Defaults to 5.0. This factor amplifies the impact of the signals in the final score calculation.

• splitter: str: Default splitter to be used if custom_splitter is not passed. Choices are from_rolling, from_custom_rolling, from_expanding, from_custom_expanding. Defaults to from_custom_rolling.

• testing_window: int: The size of the testing window. Defaults to 60.

• training_window: int: The size of the training window. Defaults to 365.

• trend_weight: float: Trend weights. Defaults to 0.3. This is the weight applied to the trend in the final score calculation.

• vol_window: int: Volatility window size. Defaults to 30. This is the window size used for calculating volatility.