Home > Research > Publications & Outputs > A Novel Forecasting Model for the Baltic Dry In...

Associated organisational units

Electronic data

  • Accepted manscruipt

    Rights statement: This is the peer reviewed version of the following article: Makridakis, S, Merikas, A, Merika, A, Tsionas, MG, Izzeldin, M. A novel forecasting model for the Baltic dry index utilizing optimal squeezing. Journal of Forecasting. 2019; 1– 13. 10.1002/for.2613 which has been published in final form at https://onlinelibrary.wiley.com/doi/full/10.1002/for.2613 This article may be used for non-commercial purposes in accordance With Wiley Terms and Conditions for self-archiving.

    Accepted author manuscript, 840 KB, PDF document

    Embargo ends: 22/05/21

    Available under license: CC BY-NC: Creative Commons Attribution-NonCommercial 4.0 International License

Links

Text available via DOI:

View graph of relations

A Novel Forecasting Model for the Baltic Dry Index Utilizing Optimal Squeezing

Research output: Contribution to journalJournal article

Published
Close
<mark>Journal publication date</mark>1/01/2020
<mark>Journal</mark>Journal of Forecasting
Issue number1
Volume39
Number of pages13
Pages (from-to)56-68
Publication statusPublished
Early online date22/05/19
Original languageEnglish

Abstract

Marine transport has grown rapidly as the result of globalization and sustainable world growth rates. Shipping market risks and uncertainty have also grown and need to be mitigated with the development of a more reliable procedure to predict changes in freight rates. In this paper, we propose a new forecasting model and apply it to the Baltic Dry Index (BDI). Such a model compresses, in an optimal way, information from the past in order to predict freight rates. To develop the forecasting model, we deploy a basic set of predictors, add lags of the BDI and introduce additional variables, in applying Bayesian compressed regression (BCR), with two important innovations. First, we include transition functions in the predictive set to capture both smooth and abrupt changes in the time path of BDI; second, we do not estimate the parameters of the transition functions, but rather embed them in the random search procedure inherent in BCR. This allows all coefficients to evolve in a time-varying manner, while searching for the best predictors within the historical set of data. The new procedures predict the BDI with considerable success.

Bibliographic note

This is the peer reviewed version of the following article: Makridakis, S, Merikas, A, Merika, A, Tsionas, MG, Izzeldin, M. A novel forecasting model for the Baltic dry index utilizing optimal squeezing. Journal of Forecasting. 2019; 1– 13. 10.1002/for.2613 which has been published in final form at https://onlinelibrary.wiley.com/doi/full/10.1002/for.2613 This article may be used for non-commercial purposes in accordance With Wiley Terms and Conditions for self-archiving.