EcoSta 2018: Registration
View Submission - EcoSta2018
A0416
Title: Exterior point algorithm for change point analysis of general time series models Authors:  Chi Tim Ng - Hang Seng University of Hong Kong (Hong Kong) [presenting]
Abstract: An efficient exterior point algorithm is proposed for smoothing and change point detection of financial time series data under the penalized likelihood approach. The proposed method has $O(n)$ computational complexity and is applicable to a broad class of time series model proposed that encompasses ARMA-GARCH as a special case. Under certain conditions, the estimated model has piecewise constant coefficients. Asymptotic properties of the penalized likelihood estimators are established. The possibility of real-time forecasting that update the prediction within $O(1)$ time upon arrival of new signal is also discussed.