The Extreme Value Approach to Uniform Fixed-k Inference in High-Frequency Data
Conference
65th ISI World Statistics Congress 2025
Format: IPS Abstract - WSC 2025
Keywords: extreme value theory, high-frequency, volatility
Session: IPS 769 - Heterogeneous Data Extremes
Tuesday 7 October 10:50 a.m. - 12:30 p.m. (Europe/Amsterdam)
Abstract
We present a unified infill asymptotic framework for inferring time-varying parameters (e.g., variance, regression coefficient, correlation) from high-frequency data using fixed-k observations in local estimation blocks. Integrating volatility coupling with extreme value theory, our approach provides uniform confidence bands of the parameter path in high-dimensional settings with a growing number of potentially overlapping blocks. Numerical analysis on high-frequency finance data is included.