The structural Theta method and its predictive performance in the M4-Competition

Giacomo Sbrana () and Andrea Silvestrini
Additional contact information
Giacomo Sbrana: NEOMA Business School

No 1457, Temi di discussione (Economic working papers) from Bank of Italy, Economic Research and International Relations Area

Abstract: The Theta method is a well-established prediction benchmark widely used in forecast competitions. Introduced more than 20 years ago, this method has received significant attention, with several authors proposing different variants to improve its performance. This paper considers the multiple sources of error version of the Theta model, belonging to the family of structural time series models, and investigates its out-of-sample forecast performance using the extensive M4-Competition dataset, which includes 100,000 time series. We compare the proposed structural Theta model against several benchmarks, including all variants of the Theta method. The results clearly demonstrate its remarkable predictive abilities as it outperforms all its variants and competitors, emerging as a solid benchmark for use in forecast competitions.

Keywords: Theta method; state-space models; Kalman filter; M4-Competition; predictive accuracy (search for similar items in EconPapers)
JEL-codes: C13 C22 (search for similar items in EconPapers)
Date: 2024-06
References: View references in EconPapers View complete reference list from CitEc
Citations:

Downloads: (external link)
https://www.bancaditalia.it/pubblicazioni/temi-dis ... 457/en_tema_1457.pdf (application/pdf)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:bdi:wptemi:td_1457_24

Access Statistics for this paper

More papers in Temi di discussione (Economic working papers) from Bank of Italy, Economic Research and International Relations Area Contact information at EDIRC.
Bibliographic data for series maintained by ().