Hybrid Forecasting Model for Carbon Price: Integrating Chronological and Weekday‐Structured Temporal Representations

ABSTRACT Accurate carbon price forecasts support allowance trading, compliance planning and market supervision. This study examines whether two temporal representations of the same carbon price series—the original chronological sequence (horizontal dimension) and five weekday‐structured subsequences (vertical dimension) obtained by regrouping Monday‐to‐Friday observations—contain complementary predictive information. Both representations are modeled within a common decomposition‐reconstruction‐forecasting framework: Dynamic time warping (DTW) diagnoses the correspondence between them, complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) and sample entropy (SE) decompose and reconstruct each series, and a leaky‐integrator echo state network (LiESN) generates the forecasts. The weekday‐specific forecasts are restored to chronological order and fused with the horizontal forecast by least squares (LS), and kernel density estimation (KDE) is used to construct prediction intervals. In the Hubei carbon market, the proposed model achieves an MAE of 0.365, a MAPE of 0.800%, an RMSE of 0.538 and an R 2 of 0.949 in the test period, reducing the MAE, MAPE, and RMSE of the chronological‐only benchmark by 2.41%, 2.20%, and 2.89%, respectively, and the Wilcoxon signed‐rank test indicates that its improvements over all benchmark models are significant at the 5% level. The model also attains the highest or tied‐highest interval coverage and the highest composite interval score. Additional experiments on the Shenzhen and Chongqing carbon markets show that the incremental gain is modest, and its statistical significance is market‐dependent: It is significant in Chongqing but not in Shenzhen. These results support the view that weekday‐structured forecasts can provide complementary predictive information beyond adjacent‐day chronological dependence, without treating the weekday sequences as independent multivariate variables or claiming a newly discovered weekday anomaly.

Authors

Institutions

Publication Details

Journal
Journal of Forecasting
Published
2026-11-13
DOI
https://doi.org/10.1002/for.70242
Primary Topic
Stock Market Forecasting Methods
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Hybrid Forecasting Model for Carbon Price: Integrating Chronological and Weekday‐Structured Temporal Representations

Yun Bai, Bo Zeng, Ziqiang Pu, Shuyun Deng et al.
Journal of Forecasting
Stock Market Forecasting Methods
article

Hybrid Forecasting Model for Carbon Price: Integrating Chronological and Weekday‐Structured Temporal Representations

Yun Bai, Bo Zeng, Ziqiang Pu, Shuyun Deng, Jingjing Xie
article en

Abstract

ABSTRACT Accurate carbon price forecasts support allowance trading, compliance planning and market supervision. This study examines whether two temporal representations of the same carbon price series—the original chronological sequence (horizontal dimension) and five weekday‐structured subsequences (vertical dimension) obtained by regrouping Monday‐to‐Friday observations—contain complementary predictive information. Both representations are modeled within a common decomposition‐reconstruction‐forecasting framework: Dynamic time warping (DTW) diagnoses the correspondence between them, complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) and sample entropy (SE) decompose and reconstruct each series, and a leaky‐integrator echo state network (LiESN) generates the forecasts. The weekday‐specific forecasts are restored to chronological order and fused with the horizontal forecast by least squares (LS), and kernel density estimation (KDE) is used to construct prediction intervals. In the Hubei carbon market, the proposed model achieves an MAE of 0.365, a MAPE of 0.800%, an RMSE of 0.538 and an R 2 of 0.949 in the test period, reducing the MAE, MAPE, and RMSE of the chronological‐only benchmark by 2.41%, 2.20%, and 2.89%, respectively, and the Wilcoxon signed‐rank test indicates that its improvements over all benchmark models are significant at the 5% level. The model also attains the highest or tied‐highest interval coverage and the highest composite interval score. Additional experiments on the Shenzhen and Chongqing carbon markets show that the incremental gain is modest, and its statistical significance is market‐dependent: It is significant in Chongqing but not in Shenzhen. These results support the view that weekday‐structured forecasts can provide complementary predictive information beyond adjacent‐day chronological dependence, without treating the weekday sequences as independent multivariate variables or claiming a newly discovered weekday anomaly.

Journal of Forecasting
Chongqing Technology and Business University (CN), Guangxi University (CN)
Stock Market Forecasting Methods
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon