大阪大学　数理・データ科学教育研究センター
Center for Mathematical Modeling and Data Science,Osaka University

Stochastic mixture covariance models

Mike K. P. So (Hong Kong University of Science and Technology)

1. ホーム
2. セミナー・集中講義
3. セミナー
4. Stochastic mixture covariance models

大阪大学 金融・保険セミナーシリーズ 第59回

Stochastic mixture covariance models

Mike K. P. So (Hong Kong University of Science and Technology)

Stochastic covariance models have been explored recently to model interdependence of assets in financial time series. This approach uses a single stochastic model to capture such dependence. However, it may not be sufficient to assume a single coherence structure at all time t. In this paper, we propose the use of a mixture formulation of stochastic covariance models to generalize the approach and offers flexibility for real data applications. Parameter estimation is performed by Bayesian analysis with MCMC sampling schemes. We conduct simulation study on three different model setups and evaluate the performance of estimation as well as model selection. We also apply our modeling methods to high-frequency stock data in Hong Kong. Model selection favors a mixture model compared to a non-mixture counterpart. In the real data study, we showcase that the mixture model is able to identify the structural change of market risk as evidenced by the drastic change of mixture proportions in time.

この研究会は、経済学研究科の経営研究会と共催です