@article{ART003280435},
author={Sang-Hyeok Lee and Changho Son},
title={Short-term Forecasting of Handysize freight rates and DNN architecture for time series classification},
journal={Journal of The Korea Society of Computer and Information},
issn={1598-849X},
year={2025},
volume={30},
number={12},
pages={113-130}
TY - JOUR
AU - Sang-Hyeok Lee
AU - Changho Son
TI - Short-term Forecasting of Handysize freight rates and DNN architecture for time series classification
JO - Journal of The Korea Society of Computer and Information
PY - 2025
VL - 30
IS - 12
PB - The Korean Society Of Computer And Information
SP - 113
EP - 130
SN - 1598-849X
AB - This study forecasts weekly changes in Handysize freight rates using a simple up-or-down prediction framework. We utilize daily Baltic Handysize Index (BHSI) data from 2006 to 2024 to construct five-day weekly patterns, subsequently predicting whether the average rate for the following week will be higher or lower than that of the current week. To accomplish this, we compare a comprehensive set of models, including 48 compact deep neural networks—comprising multilayer perceptrons, fully convolutional, and residual architectures—as well as standard benchmarks such as Bi-LSTM, Transformer, support vector machine, and random forest classifiers. Out-of-time tests conducted under various market conditions demonstrate that a residual network with moderate sensitivity to intra-week ordering delivers the most accurate and stable forecasts, yielding well-calibrated probabilities. These findings indicate that weekly freight patterns encompass exploitable directional information and that the proposed residual network can function as an effective tool for chartering decisions, freight hedging, and market monitoring within the dry bulk shipping industry.
KW - Residual network;Weekly freight-rate forecasting;Baltic Handysize Index;;Time series classification;Deep neural networks;Dry bulk shipping market;;Probability calibration;Out-of-time validation
DO -
UR -
ER -
Sang-Hyeok Lee and Changho Son. (2025). Short-term Forecasting of Handysize freight rates and DNN architecture for time series classification. Journal of The Korea Society of Computer and Information, 30(12), 113-130.
Sang-Hyeok Lee and Changho Son. 2025, "Short-term Forecasting of Handysize freight rates and DNN architecture for time series classification", Journal of The Korea Society of Computer and Information, vol.30, no.12 pp.113-130.
Sang-Hyeok Lee, Changho Son "Short-term Forecasting of Handysize freight rates and DNN architecture for time series classification" Journal of The Korea Society of Computer and Information 30.12 pp.113-130 (2025) : 113.
Sang-Hyeok Lee, Changho Son. Short-term Forecasting of Handysize freight rates and DNN architecture for time series classification. 2025; 30(12), 113-130.
Sang-Hyeok Lee and Changho Son. "Short-term Forecasting of Handysize freight rates and DNN architecture for time series classification" Journal of The Korea Society of Computer and Information 30, no.12 (2025) : 113-130.
Sang-Hyeok Lee; Changho Son. Short-term Forecasting of Handysize freight rates and DNN architecture for time series classification. Journal of The Korea Society of Computer and Information, 30(12), 113-130.
Sang-Hyeok Lee; Changho Son. Short-term Forecasting of Handysize freight rates and DNN architecture for time series classification. Journal of The Korea Society of Computer and Information. 2025; 30(12) 113-130.
Sang-Hyeok Lee, Changho Son. Short-term Forecasting of Handysize freight rates and DNN architecture for time series classification. 2025; 30(12), 113-130.
Sang-Hyeok Lee and Changho Son. "Short-term Forecasting of Handysize freight rates and DNN architecture for time series classification" Journal of The Korea Society of Computer and Information 30, no.12 (2025) : 113-130.
