A1478
Title: Detecting structural shifts in interval-based time series
Authors: Li-Hsien Sun - National Central University (Taiwan) [presenting]
Chi-Yang Chiu - University of Tennessee Health Science Center (United States)
Abstract: Multiple change-points analysis for interval-valued time series data is addressed through a proposed model containing daily opening (O), up (U), low (L), and closing (C) values, based on the log-transformed geometric Brownian motion model rather than traditional opening and closing values models. The joint distribution for these interval-valued observations is obtained using the reflection principle and Girsanov's theorem. Change-points are identified by the MLE method and PELT algorithm. Performance of the model is investigated through extensive simulations and real data analysis using S\&P500 returns during the 2020 COVID-19 pandemic. Results demonstrate that the proposed OULC model consistently outperforms the traditional OC model in offering more accurate and reliable change-point detection.
