Probabilistic framework for modeling event shocks to financial time series.

In financial market, certain types of stochastic events are intrinsically impactful to the prediction of financial times series, such as stock return, while few existing research attempts have been made to incorporate stochastic event modeling to time series modeling in a principled way. In this paper, we present a pioneering study that fills this gap. In particular, we introduce a generic probabilistic model that captures 1) the inter-dependencies among stochastic events, and 2) the impact of these events on time series. To this end, we extend multivariate Hawkes process (MHP) and proximal graphical event model (PGEM) and apply this framework to modeling two financial events, companies' quarterly revenue releases and updates of consensus prediction of quarterly revenue, and their impacts on the mean and correlation structures of future stock return. Our model not only improves prediction of financial time series, but also promotes AI trust for finance by revealing the causal relationship among the events. Extensive experimental results based on real financial market data validate the effectiveness of our models in learning event impact and improving investment decision by incorporating stochastic event impacts.