Generalized Autoregressive Score Models R package GAS version 0.4.3 Simulation, estimation and forecasting of univariate and multivariate generalized autoregressive score models. Time-varying distributional parameters are updated using the score of the conditional distribution, with alternative score scaling specifications. Provides maximum likelihood estimation, filtering, density forecasting and rolling forecast evaluation for a range of conditional distributions. Includes diagnostic tools and methods for evaluating Value-at-Risk and Expected Shortfall forecasts. The models and their applications are described in Ardia et al. (2019) <doi:10.18637/jss.v088.i06>. Software Depends: R (>= 3.4.0), methods Imports: Rcpp (>= 0.12.2), Rsolnp, MASS, xts, numDeriv, zoo, cubature LinkingTo: Rcpp, RcppArmadillo Suggests: testthat Leopoldo Catania <leopoldo.catania@econ.au.dk> Comprehensive R Archive Network (CRAN) Leopoldo Catania [aut, cre] (ORCID: <https://orcid.org/0000-0002-0981-1921>), David Ardia [ctb] (ORCID: <https://orcid.org/0000-0003-2823-782X>), Kris Boudt [ctb] (ORCID: <https://orcid.org/0000-0002-1000-5142>) GPL-3 2026-10-30 application/tgz https://CRAN.R-project.org/package=GAS doi:10.32614/CRAN.package.GAS