Gradient Boosting GARCH and Neural Networks for Time Series Prediction
This work develops and evaluates new algorithms based on neural networks and boosting techniques, designed to model and predict heteroskedastic time series. The main novel elements of these new algorithms are as follows: a) in regard to neural networks, the simultaneous estimation of conditional mean and volatility through the likelihood maximization; b) in regard to boosting, its simultaneous application to trend and volatility components of the likelihood, and the use of likelihood-based models (e.g. GARCH) as the base hypothesis rather than gradient fitting techniques using least squares. The behavior of the proposed algorithms is evaluated over simulated data, resulting in frequent and significant improvements in relation to the ARMA-GARCH models.
Keywordsgradient boosting GARCH ARMA neural networks time series heteroskedasticity
Unable to display preview. Download preview PDF.
- 1.Audrino F, Bühlmann P (2002) Volatility estimation with functional gradient descent for very high dimensional financial time series. J Comp Finance 6:65–89Google Scholar
- 6.Foresee FD, Hagan T (1997) Gauss-Newton approximation to Bayesian regularization. In: Proc IEEE Int Joint Conf Neural Networks, Houston, TX, USA. IEEE, Piscataway, pp 1930–1935Google Scholar
- 10.Medeiros MC, Teräsvirta T, Rech G (2002) Building neural networks models for time series: a statistical approach. Tech Report 461, Stockholm School of Economics, StockholmGoogle Scholar
- 14.Schapire RE (1990) The strength of weak learnability. Mach Learn 5:197–227Google Scholar
- 17.Tsay RS (2001) Analysis of financial time series. John Wiley and Sons, HobokenGoogle Scholar
- 18.Weigend AS, Gershenfeld NA (1993) Time series prediction: forecasting the future and understanding the past. Perseus Books, New YorkGoogle Scholar