We forecast daily realized volatilities with linear and nonlinear models and evaluate the benefits of bootstrap aggregation (bagging) in producing more precise forecasts. We consider the linear autoregressive (AR) model, the Heterogeneous Autoregressive model (HAR), and a non-linear HAR model based on a neural network specification that allows for logistic transition effects (NNHAR). The models and the bagging schemes are applied to the realized volatility time series of the S&P500 index from 3-Jan-2000 through 30-Dec-2005. Our main findings are: (1) For the HAR model, bagging successfully averages over the randomness of variable selection; however, when the NN model is considered, there is no clear benefit from using bagging; (2) including past returns in the models improves the forecast precision; and (3) the NNHAR model outperforms the linear alternatives.
Download Info
To download:
If you experience problems downloading a file, check if you have the
proper application to
view it first. Information about this may be contained
in the File-Format links below. In case of further problems read
the IDEAS help
page. Note that these files are not on the IDEAS
site. Please be patient as the files may be large.
Publisher Info
Paper provided by Department of Economics PUC-Rio (Brazil) in its series Textos para discussão with number
547.
Cited by: (explanations, Please report citation or reference errors to , or , if you are the registered author of the cited work, log in to your RePEc Author Service profile, click on "citations" and make appropriate adjustments.)