A0183
Title: Detection and estimation of changepoints within time-dependent functional profiles
Authors: Matus Maciak - Charles University (Czech Republic) [presenting]
Sebastiano Vitali - University of Bergamo (Italy)
Abstract: The problem of analyzing financial markets properly relies, among others, on the ability to detect, estimate, and understand different types of structural changes---changepoints. We particularly focus on changes within specific time-dependent functional profiles obtained from the observed options' implied volatility (IV) smiles, and we discuss two different but mutually related approaches: Firstly, consistent statistical tests for detecting significant changepoints are proposed and investigated under different theoretical assumptions and different practical scenarios. Second, an overall stochastic model is postulated, and the unknown (sparse) changepoints are estimated using a regularized quantile estimation framework---all within a model that fully complies with the theory on arbitrage-free markets. Theoretical and empirical aspects are both addressed in detail, and some finite sample performance is illustrated in terms of a simulation study and real data examples.
