Dynamical Products of Experts for Modeling Financial Time Series

Predicting the "Value at Risk" of a portfolio of stocks is of great significance in quantitative finance. We introduce a new class models, "dynamical products of experts" that treats the latent process over volatilities as an inverse Gamma process. We show that our multivariate volatility models significantly outperform all related Garch and stochastic volatility models which are in popular use in the quantitative finance community.

• Type: Conference
• Year: 2010
• Where: ICML
• Authors: Yutian Chen, Max Welling
• Keywords: ICML 2010, Machine Learning, Multivariate Volatility Models, Quantitative Finance, Volatility Models