论文信息 - Augmented Functional Time Series Representation and Forecasting with Gaussian Processes

Augmented Functional Time Series Representation and Forecasting with Gaussian Processes

We introduce a functional representation of time series which allows forecasts to be performed over an unspecified horizon with progressively-revealed information sets. By virtue of using Gaussian processes, a complete covariance matrix between forecasts at several time-steps is available. This information is put to use in an application to actively trade price spreads between commodity futures contracts. The approach delivers impressive out-of-sample risk-adjusted returns after transaction costs on a portfolio of 30 spreads.

Nicolas Chapados | Yoshua Bengio

[1] A. O'Hagan,et al. Curve Fitting and Optimal Design for Prediction , 1978 .

[2] J. Hull. Options, Futures, and Other Derivatives , 1989 .

[3] Heekuck Oh,et al. Neural Networks for Pattern Recognition , 1993, Adv. Comput..

[4] C. Rasmussen,et al. Gaussian Process Priors with Uncertain Inputs - Application to Multiple-Step Ahead Time Series Forecasting , 2002, NIPS.

[5] Nicolas Chapados,et al. Cost functions and model combination for VaR-based asset allocation using neural networks , 2001, IEEE Trans. Neural Networks.

[6] William N. Goetzmann,et al. Active Portfolio Management , 1999 .

[7] Volker Tresp,et al. A Bayesian Committee Machine , 2000, Neural Computation.

[8] Carl E. Rasmussen,et al. A Unifying View of Sparse Approximate Gaussian Process Regression , 2005, J. Mach. Learn. Res..

[9] Carl E. Rasmussen,et al. In Advances in Neural Information Processing Systems , 2011 .

[10] Jonathan D. Cryer,et al. Time Series Analysis , 1986, Encyclopedia of Big Data.

[11] F. Diebold,et al. Comparing Predictive Accuracy , 1994, Business Cycles.

[12] Carl E. Rasmussen,et al. Gaussian processes for machine learning , 2005, Adaptive computation and machine learning.