We consider a state space model approach forhigh frequency financial data analysis. An expectationmaximization(EM) algorithm is developed for estimatingthe integrated covariance matrix of the assets. The statespace model with the EM algorithm can handle noisy financialdata with correlated microstructure noises. Difficultydue to asynchronous and irregularly spaced trading data ofmultiple assets can be naturally overcome by consideringthe problem in a scenario with missing data. Since the statespace model approach requires no data synchronization, norecord in the financial data is deleted so that it efficientlyincorporates information from all observations. Empiricaldata analysis supports the general specification of the statespace model, and simulations confirm the efficiency
EM algorithm, High frequency data, Integrated covariance matrix, Kalman Filter, Microstructure noise, Missing data, Quasi-maximum likelihood, State Space Model
Econometrics | Economic Theory
Statistics and Its Interface
Liu, Cheng and TANG, Cheng Yong.
A state space model approach to integrated covariance matrix estimation with high frequency data. (2013). Statistics and Its Interface. 6, (4), 463-475. Research Collection Lee Kong Chian School Of Business.
Available at: http://ink.library.smu.edu.sg/lkcsb_research/5603
Creative Commons License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.