A0412
Title: Interval-based time series analysis: Detecting structural shifts and estimating change-points
Authors: Li-Hsien Sun - National Central University (Taiwan) [presenting]
Abstract: A novel method is presented for detecting structural shifts within interval-based time series data and accurately estimating the change point. The approach involves developing a financial time series model that considers daily maximum, minimum, and terminal values based on the geometric Brownian motion model. The proposed method is then applied to the financial time series where the data is emphasized by the significance of intra-daily information, including maximum and minimum prices, while traditional finance models primarily focus on the daily closing price given the open price. The likelihood function and corresponding maximum likelihood estimates (MLEs) are derived using the Girsanov theorem and the Newton-Raphson (NR) algorithm. Through extensive simulations, the effectiveness of the proposed approach is thoroughly evaluated. Furthermore, empirical studies using real stock return data (S\&P 500 index) from two critical periods, the 2008 financial crisis, the COVID-19 pandemic in 2020, and the Russo-Ukrainian War in 2022, allow assessing its performance robustly.
