Dynamic Covariance Models for Multivariate Financial Time Series

By Yue Wu, José Miguel Hernández-Lobato, Zoubin Ghahramani. 2013

50 citations recorded by the source corpus.

The accurate prediction of time-changing covariances is an important problem in the modeling of multivariate financial data. However, some of the most popular models suffer from a) overfitting problems and multiple local optima, b) failure to capture shifts in market conditions and c) large computational costs. To address these problems we introduce a novel dynamic model for time-changing covariances. Over-fitting and local optima are avoided by following a Bayesian approach instead of computing point estimates. Changes in market conditions are captured by assuming a diffusion process in parameter values, and finally computationally efficient and scalable inference is performed using particle filters. Experiments with financial data show excellent performance of the proposed method with respect to current standard models.

Subfields: Economics and Econometrics, Finance, Signal Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Economics and Econometrics

1. S. Mei
2. Min Liu
3. Jun Wang
4. Sivagowry Sriananthakumar
5. Renzhuo Wan
6. Seema W. Narayan
7. Fan Yang
8. S. Islam

In Finance

1. Sivagowry Sriananthakumar
2. Seema W. Narayan
3. Fan Yang
4. S. Islam
5. Paresh Kumar Narayan

In Signal Processing

1. Fan Yang

Citations in the corpus

• Stock market integration of emerging Asian economies: Patterns and causes (2014, background): “One issue with the 1 Time-varying covariances are not considered because their estimation involves some practical difficulties such as a) overfitting problems andmultiple local optima; b) failure to capture shifts in market conditions; and c) large computational costs (Wu et al., 2013).”
• Multivariate Temporal Convolutional Network: A Deep Neural Networks Approach for Multivariate Time Series Forecasting (2019, background): “With the explosive growth of Internet of Things (IoT) applications and big data, multivariate time series is becoming ubiquitous in many fields, e.g., aerology [1], meteorology [2], environment [3], multimedia [4], power energy [5], finance [6], and transportation [7].”