Day-ahead crude oil price forecasting using a novel morphological component analysis based model

Area of Science:

• Econometrics
• Financial Modeling
• Data Science

Background:

• Crude oil price movements exhibit complex nonlinear and dynamic behavior, making accurate forecasting a significant challenge.
• Recent research highlights the importance of multiscale data characteristics as a stylized fact in price movements.
• Incorporating these multiscale characteristics can substantially enhance predictive model performance.

Purpose of the Study:

• To propose a novel hybrid methodology for modeling the multiscale heterogeneous characteristics of crude oil price movements.
• To improve the accuracy of crude oil price forecasting by accounting for complex data structures.
• To offer economic interpretations of the heterogeneous market microstructure.

Main Methods:

• Development of a hybrid methodology based on morphological component analysis.
• Modeling multiscale heterogeneous characteristics within the time scale domain.
• Empirical validation using benchmark crude oil markets.

Main Results:

• Empirical studies confirmed the existence of a multiscale heterogeneous microdata structure in crude oil markets.
• The proposed algorithm demonstrated significant performance improvement compared to benchmark models (random walk, ARMA, SVR).
• The methodology effectively incorporates multiscale heterogeneous data characteristics for enhanced modeling.

Conclusions:

• The novel morphological component analysis-based hybrid methodology successfully models multiscale heterogeneous characteristics in crude oil prices.
• Accounting for these characteristics leads to superior forecasting performance.
• The study provides valuable insights into market microstructure with economic relevance.