1,721,052 research outputs found
Nonlinear time series models with switching structure: a comparison of their forecast performances
No full textThreshold Vector ARMA Forecasts under General Loss Functions
No full textIn linear time series analysis point forecasts are based on the minimization of a square loss function which allows to obtain the conditional expectation as optimal predictor.
When the data generating process is nonlinear, the forecaster should use general loss functions that are able to take into account some features of the underlying process.
The introduction of “general loss functions” has been widely investigated in univariate time series domain (see among the others Christoffersen and Diebold (1996, 1997), Granger (1999), Patton and Timmermann (2007a, 2007b, 2010)) whereas, in our knowledge, in multivariate domain the use of this kind of functions has been only marginally explored in Alp and Demetrescu (2010) and Komunjer and Owyang (2010).
Starting from these results, in our contribution we present a new class of multivariate nonlinear time series models called Threshold Vector ARMA (TVARMA) that generalizes the Threshold ARMA model proposed in Tong (1983). After the presentation of some properties of the TVARMA model, we propose the use of asymmetric loss functions to generate forecasts. In more detail we show that these functions are able to catch some features of the model that are completely neglected by square loss functions
Multi-step forecasts from threshold ARMA models using asymmetric loss functions
No full textNonlinear prediction, General loss functions, SETARMA, Linex,
Non-Linear Dynamics and Evaluation of Forecasts using High-Frequency Time Series
No full textIn the present paper we evaluate the performance of a non linear parametric model in forecasting high-frequency data. In particular we consider the TAR-ARCH model (Li and Lam ,1995) to fit and forecast the daily and 5-minute returns of the Mibtel Stock Index
Predictor distribution and forecast accuracy of threshold models
No full textIn the present paper the predictor distribution of a SETAR (Self Exciting Threshold Autoregressive) model (Tong and Lim, 1980) has been investigated when the lead time is greater than the threshold delay. After a brief presentation of the model under study, some relevant aspects of the density forecasts are shown highlighting how they can be used to generate more accurate predictions and to estimate an approximation of the probability density function of the SETAR predictors. The performances of competing predictors have been evaluated through a simulation study and an application to financial market data of the daily Nikkey 300 stock market returns
Forecast density of regimes switching conditional heteroskedastic models
No full textIn the present paper the accuracy of multi-step ahead predictors has been evaluated through the forecast densities of a selection of nonlinear time series structures which present conditional variance changing over time. The forecast densities and the forecast regions have been estimated using a Monte Carlo simulation procedure. The relevance of the estimated coefficients on the amplitude of the forecast regions has been investigated and the role of the model intercepts on the density shape of the regime switching models have been examined
Predictive Distributions of Nonlinear Time Series Models
No full textWorking Paper 3.102, Dipartimento di Scienze Economich
- …
