Abstract
We analyze the complete tick-level stock trading records at the Taiwan Stock Exchange and explore what factors are principally associated with stock price jumps in high frequency. Among the potential candidate variables suggested in the literature, liquidity proxies appear to be primarily associated with signed jumps in high frequency. The results from the least absolute shrinkage and selection operator (LASSO), the elastic net method, and principal component analysis further show that liquidity issues are more important than information or sentiment in understanding sudden and discontinuous price innovations to financial assets in high frequency.
Original language | English |
---|---|
Article number | 101602 |
Journal | Pacific Basin Finance Journal |
Volume | 68 |
DOIs | |
Publication status | Published - 2021 Sept |
Keywords
- Big data
- Herding
- Information asymmetry
- Liquidity
- Machine learning
- Signed jumps
ASJC Scopus subject areas
- Finance
- Economics and Econometrics