High frequency financial time series analysis using partially quantified PCA
Eun ji Han, Jae Eun Yoon, Sun Young Hwang · Korean Journal of Applied Statistics · 2025
This article is concerned with high frequency financial time series analysis for which partially-quantified principal component analysis (PCA, for short) is exploited.Partially-quantified PCA is useful especially when the first principal component is subjectively given, for a practical purpose, prior to the usual PCA analysis.High frequency financial time series consists of a lot of intraday returns and thus partially-quantified PCA may help provide a successful dimension reduction for the data.Interesting applications are made to domestic post-Covid-19 financial data.Specifically, four sets of one-minute high frequency financial data including KOSPI (Korea stock prices index) spanning from January 2022 to July 2024 are analyzed via partially-quantified PCA to illustrate low-dimensional data reduction for the intraday returns.