ABSTRACT This paper employs Principal Component Analysis (PCA) to identify time‐varying latent factors from high‐frequency (HF) stock return data in the Chinese market. We find these factors effectively explain industry portfolio returns during both trading and non‐trading hours. Moreover, long‐short portfolios constructed based on exposures to the HF PCA factors yield significant abnormal returns. We provide economic interpretations of the HF PCA factors by exploring their relationships with macroeconomic variables through LASSO regression. Various economic variables, such as the Unemployment Rate, Growth of Industrial Added Value, and Macroeconomic Consensus Index, are identified to have strong correlation with these statistical factors.
Zhu et al. (Mon,) studied this question.