EcoSta 2017: Start Registration
View Submission - EcoSta2017
A0667
Title: Bayesian analysis of generalized long-memory stochastic volatility Authors:  Alex Gonzaga - University of the Philippines Manila (Philippines) [presenting]
Abstract: A Bayesian approach is proposed for estimating the parameters and future values of the Generalized Long-memory Stochastic Volatility (GLMSV) model will be predicted by utilizing the approximate likelihood function of discrete wavelet packet coefficients. This provides an alternative method incorporating prior information about the model parameters and utilizing the decorrelating property of wavelet packet transform of the model. This simplifies the variance-covariance matrix of the model by approximately decorrelating wavelet coefficients within and across scales. This approximation does not depend on the signal, but the length of the wavelet filter, which is under the control of the analyst. It allows for a computationally efficient sampling from the posterior density of the parameters in a Bayesian approach for estimating and predicting future values. The proposed method is then applied in the analysis of volatility of financial time series, such as the intraday volatility of stock prices.