A Study of Indian Stock Market using Matrix Variate Dynamic Model

Our aim is to study Indian stock market data using matrix variate dynamic model (MVDM). We need a model which preserves non-linear structure and the complexity of stock market data. MVDM has some structure preserving property which helps in understanding behaviour of stock data. We model MVDM by taking latent and observed data as a matrix variate with Gaussian noise. We use expectation maximization iterative techniques for parameter estimation. MVDM prediction accuracy is more except during major political events and COVID-19 period.