Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Multistep Brent Oil Price Forecasting with a Multi-Aspect Meta-heuristic Optimization and Ensemble Deep Learning Model

Mohammed Alruqimi, Luca Di Persio

TL;DR

This study tackles the challenge of accurate multi-step Brent crude oil price forecasting by introducing a hybrid ensemble that leverages the Grey Wolf Optimizer (GWO) across four stages: feature selection, data preparation (sliding window), neural network hyperparameter tuning, and forecast blending. The ensemble comprises five architectures (Bi-LSTM, Bi-GRU, CNN-Bi-LSTM, CNN-Bi-LSTM-att, and encoder-decoder-Bi-LSTM), trained individually and then fused with GWO-optimised weights. On Brent price data from 2012–2021 with USDX and SENT as exogenous features, the GWO-driven approach achieves an ensemble $MSE$ of $0.000127$, with the best individual model (SENT-Bi-GRU) reaching $MSE = 0.000137$. The results demonstrate that multi-aspect optimization and conditional forecast blending markedly improve forecasting performance, offering practical benefits for energy risk management and trading strategies. Future work could explore broader parameter spaces and computationally efficient search strategies to further enhance scalability.

Abstract

Accurate crude oil price forecasting is crucial for various economic activities, including energy trading, risk management, and investment planning. Although deep learning models have emerged as powerful tools for crude oil price forecasting, achieving accurate forecasts remains challenging. Deep learning models' performance is heavily influenced by hyperparameters tuning, and they are expected to perform differently under various circumstances. Furthermore, price volatility is also sensitive to external factors such as world events. To address these limitations, we propose a hybrid approach that integrates metaheuristic optimization with an ensemble of five widely used neural network architectures for time series forecasting. Unlike existing methods that apply metaheuristics to optimise hyperparameters within the neural network architecture, we exploit the GWO metaheuristic optimiser at four levels: feature selection, data preparation, model training, and forecast blending. The proposed approach has been evaluated for forecasting three-ahead days using real-world Brent crude oil price data, and the obtained results demonstrate that the proposed approach improves the forecasting performance measured using various benchmarks, achieving 0.000127 of MSE.

Multistep Brent Oil Price Forecasting with a Multi-Aspect Meta-heuristic Optimization and Ensemble Deep Learning Model

TL;DR

This study tackles the challenge of accurate multi-step Brent crude oil price forecasting by introducing a hybrid ensemble that leverages the Grey Wolf Optimizer (GWO) across four stages: feature selection, data preparation (sliding window), neural network hyperparameter tuning, and forecast blending. The ensemble comprises five architectures (Bi-LSTM, Bi-GRU, CNN-Bi-LSTM, CNN-Bi-LSTM-att, and encoder-decoder-Bi-LSTM), trained individually and then fused with GWO-optimised weights. On Brent price data from 2012–2021 with USDX and SENT as exogenous features, the GWO-driven approach achieves an ensemble of , with the best individual model (SENT-Bi-GRU) reaching . The results demonstrate that multi-aspect optimization and conditional forecast blending markedly improve forecasting performance, offering practical benefits for energy risk management and trading strategies. Future work could explore broader parameter spaces and computationally efficient search strategies to further enhance scalability.

Abstract

Accurate crude oil price forecasting is crucial for various economic activities, including energy trading, risk management, and investment planning. Although deep learning models have emerged as powerful tools for crude oil price forecasting, achieving accurate forecasts remains challenging. Deep learning models' performance is heavily influenced by hyperparameters tuning, and they are expected to perform differently under various circumstances. Furthermore, price volatility is also sensitive to external factors such as world events. To address these limitations, we propose a hybrid approach that integrates metaheuristic optimization with an ensemble of five widely used neural network architectures for time series forecasting. Unlike existing methods that apply metaheuristics to optimise hyperparameters within the neural network architecture, we exploit the GWO metaheuristic optimiser at four levels: feature selection, data preparation, model training, and forecast blending. The proposed approach has been evaluated for forecasting three-ahead days using real-world Brent crude oil price data, and the obtained results demonstrate that the proposed approach improves the forecasting performance measured using various benchmarks, achieving 0.000127 of MSE.
Paper Structure (22 sections, 3 equations, 6 figures, 4 tables)

This paper contains 22 sections, 3 equations, 6 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Litrature Review
  3. Preliminaries
  4. Metaheuristic optimisation for machine learning models
  5. Grey Wolf Optimizer (GWO)
  6. Proposed approach
  7. Individual models
  8. GWO ensemble model
  9. GWO optimisation
  10. Feature selection
  11. Sliding window size
  12. Neural networks hyperparameters tuning
  13. Blending of final forecasts
  14. Experiments
  15. Dataset and input features
  16. ...and 7 more sections

Figures (6)

  • Figure 1: Typical wolf-pack hunting behaviours: (A) chasing, approaching, and tracking prey (B–D) pursuing, harassing, and encircling (E) stationary situation and attack MURO2011192.
  • Figure 2: Proposed Approach
  • Figure 3: Brent oil price trend from 2012 to 2021
  • Figure 4: Historical data of Brent oil price, US dollar index (USDX), and the sentimental score (SENT)
  • Figure 5: Actual vs. predicted results of the examined models: (a). Ensemble model, (b). SENT-Bi-GRU, (c). SENT-Bi-LSTM, (d). SENT-CNN-Bi-LSTM, (e). SENT-CNN-Bi-LSTM-att, (f). SENT-USD-Encoder-Decoder-Bi-LSTM
  • ...and 1 more figures