forecast_svp: Multi-Step Ahead Volatility Forecast

In wARMASVp: Winsorized ARMA Estimation for Higher-Order Stochastic Volatility Models

Multi-Step Ahead Volatility Forecast

Description

Applies Kalman filtering/smoothing to an estimated SV(p) model and produces multi-step ahead volatility forecasts with uncertainty quantification.

Usage

forecast_svp(
  object,
  H = 1,
  output = c("log-variance", "variance", "volatility"),
  filter_method = "corrected",
  proxy = c("bayes_optimal", "u"),
  K = 7,
  M = 1000,
  seed = 42,
  del = 1e-10
)

Arguments

object	An "svp", "svp_t", or "svp_ged" object from svp.
H	Integer. Maximum forecast horizon. Default 1.
output	Character. Primary output scale: "log-variance" (default, native log-volatility w_h), "variance" (conditional variance \sigma^2_{T+h|T}), or "volatility" (conditional std dev \sigma_{T+h|T}). All three are always computed and stored; this controls which is used by print and plot methods.
filter_method	Character. Filter method: "corrected" (default), "mixture" (GMKF), or "particle" (BPF).
proxy	Character. Leverage proxy for the filter and the h=1 forecast shift. "bayes_optimal" (default) uses the posterior mean E[\zeta_{t-1} \mid u_{t-1}] for Student-t leverage; "u" reproduces the paper-faithful proxy of Remark 3.5 (\hat{z}_{t-1} = \hat{u}_{t-1}). Has no effect for Gaussian or GED leverage. See filter_svp for details.
K	Integer. Number of mixture components for GMKF. Default 7.
M	Integer. Number of particles for BPF. Default 1000.
seed	Integer. Random seed for BPF. Default 42.
del	Numeric. Small constant for log transformation. Default 1e-10.

Value

An object of class "svp_forecast", a list containing:

w_forecasted

Primary forecast (scale determined by output).

log_var_forecast

Log-volatility forecasts w_{T+h|T}.

var_forecast

Conditional variance forecasts \sigma^2_{T+h|T}.

vol_forecast

Conditional volatility forecasts \sigma_{T+h|T}.

P_forecast

Forecast MSE P_{T+h|T} for each horizon.

w_estimated

Filtered log-volatility.

w_smoothed

Smoothed log-volatility.

zt

Filtered standardized residuals.

zt_smoothed

Smoothed standardized residuals.

ys

Demeaned log-squared returns.

mdl

The estimated model object.

H

The forecast horizon.

output

The chosen output scale.

filter_output

The "svp_filter" object from filtering.

Examples

sim <- sim_svp(1000, phi = 0.95, sigy = 1, sigv = 0.2,
               leverage = TRUE, rho = -0.3)
fit <- svp(sim$y, p = 1, leverage = TRUE)
fc <- forecast_svp(fit, H = 10)
plot(fc)

wARMASVp documentation built on May 15, 2026, 5:07 p.m.