arch.univariate.ARCH.forecast¶

ARCH.forecast(parameters: ndarray | Series, resids: ndarray[Any, dtype[float64]], backcast: float | ndarray[Any, dtype[float64]], var_bounds: ndarray[Any, dtype[float64]], start: int | None = None, horizon: int = 1, method: 'analytic' | 'simulation' | 'bootstrap' = 'analytic', simulations: int = 1000, rng: Callable[[int | tuple[int, ...]], ndarray[Any, dtype[float64]]] | None = None, random_state: RandomState | None = None) → VarianceForecast¶

Forecast volatility from the model

Parameters:¶

parameters: ndarray | Series¶: Parameters required to forecast the volatility model
resids: ndarray[Any, dtype[float64]]¶: Residuals to use in the recursion
backcast: float | ndarray[Any, dtype[float64]]¶: Value to use when initializing the recursion
var_bounds: ndarray[Any, dtype[float64]]¶: Array containing columns of lower and upper bounds
start: int | None = None¶: Index of the first observation to use as the starting point for the forecast. Default is len(resids).
horizon: int = 1¶: Forecast horizon. Must be 1 or larger. Forecasts are produced for horizons in [1, horizon].
method: 'analytic' | 'simulation' | 'bootstrap' = 'analytic'¶: Method to use when producing the forecast. The default is analytic.
simulations: int = 1000¶: Number of simulations to run when computing the forecast using either simulation or bootstrap.
rng: Callable[[int | tuple[int, ...]], ndarray[Any, dtype[float64]]] | None = None¶: Callable random number generator required if method is ‘simulation’. Must take a single shape input and return random samples numbers with that shape.
random_state: RandomState | None = None¶: NumPy RandomState instance to use when method is ‘bootstrap’

Returns:¶

forecasts – Class containing the variance forecasts, and, if using simulation or bootstrap, the simulated paths.

Return type:¶

arch.univariate.volatility.VarianceForecast

Raises:¶

Notes

The analytic method is not supported for all models. Attempting to use this method when not available will raise a ValueError.