Integrated nested Laplace approximations for threshold stochastic volatility models

• Implementation of INLA to threshold stochastic volatility models is reliable for the estimation TSV parameters in finite samples very fast seem improve forecasting equity, commodity and cryptocurrency returns The aim implement integrated nested Laplace approximations (INLA), known be efficient, estimating (TSV) model. replaces Markov chain Monte Carlo (MCMC) simulations with accurate deterministic approximations. Weakly informative proper priors are used, as well Penalizing Complexity (PC) priors. simulation results favor use PC priors, specially when sample size varies from small moderate. For these sizes, provide more estimates model parameters. However, increases, both types lead similar method applied six series returns, including stock market, its performance assessed, by means in-sample out-of-sample approaches; one-day-ahead volatilities also carried out. empirical support that generally fits best most times ranks first terms volatility, compared symmetric