High-frequency financial big data analysis

Abstract

The wide availability of financial big data has significantly increased our ability to understand social and economic phenomena. However, utilizing big data often introduces new challenges due to its complex structure and high-dimensionality. For instance, in working with high-frequency data, we encounter complexities caused by microstructure noise and heavy-tailed distributions. Similarly, with high-dimensional data, the curse of dimensionality becomes a critical issue to address. Therefore, it is important to develop effective and efficient estimation methods for big data. In this thesis, we develop the well-performing nonparametric estimation methods and parametric models for the financial big data.