Research Speaker: Lyubov Doroshenko – PhD Student and SAMSI Visitor, La Sapienza University of Rome
Supervised by: Liseo Brunero, Christian Macaro
Title: A Mixture of Heterogeneous Models with Time Dependent Weights
Understanding stock market volatility is a major task for market analysts, policy makers, economists and investors. However, inference in financial and economic models can be challenging due to the fact that an explicit dependence order between observations is added: a time dimension. Some of the existing approaches aim to address these challenges by using ARMA, GARCH, Dynamic Linear Models and many others. In this work, we provide an alternative way to model and predict these types of data with the help of a mixture of heterogeneous models with mixing weights characterized by an autoregressive structure. In comparison to the static mixture, the models we introduce are based on time-dependent weights which allows to learn if the data generating mechanism changes over time. The resulting dynamic mixtures aim to model the composition of the stock market data. A subjective Bayesian approach is adopted and the Metropolis-Hastings within Gibbs sampling technique is used. Through extensive analysis in both real and simulated data settings, we show all the benefits that our dynamic mixture model has over its static counterpart. We illustrate this performance in the context of the stock market expectation of a 30-day forward-looking volatility expressed by the volatility index VIX.