Research On Crude Oil Price Prediction Based On MEEMD Algorithm And OPEC Policy

As an important energy source and raw material for modern industry,crude oil has a very wide range of applications in social and economic development,and thus its price fluctuations can affect multiple dimensions of national economic development,thus having a significant impact on the global economy.Crude oil price fluctuations are influenced by many complex factors,such as:supply and demand,inventories,economic,financial,geopolitical and extreme events.Crude oil price fluctuations have obvious non-linear characteristics,and due to this complex oil price structure,it makes oil price and volatility forecasting a difficult task.In a time of resource scarcity,crude oil is not only a necessary resource for production,but also a profitable commodity.Accurate forecasting of crude oil price trend is of great importance to policy makers and investors for risk management and portfolio development.Therefore,how to improve the accuracy of crude oil price forecasting has become a hot topic of academic research at present.In order to better predict crude oil prices,this paper uses decomposition integration algorithm to enhance the prediction accuracy of the underlying model on the one hand,and studies the mechanism of crude oil policies of oil-exporting countries on oil prices on the other hand,so as to assist the prediction model in forecasting crude oil prices.The main points of this article are summarized as follows:(1)Research on Crude Oil Price Forecasting Based on Decomposition Ensemble Algorithm.International crude oil prices receive crossover effects from multiple factors and thus exhibit non-linearity and dynamism.Using single forecasting models such as autoregressive moving average model(ARIMA),linear regression model(LR),extreme learning machine model(ELM)and support vector regression model(SVR)performs poorly in forecasting such complex data with significant nonlinear characteristics.To cope with the problem of low forecasting accuracy due to the nonlinearity of crude oil series,this paper uses the idea of decomposition integration for forecasting crude oil prices.Specifically,we use a decomposition algorithm to decompose crude oil prices into different modal components,where each modal component represents different frequencies of the crude oil series,which have smaller nonlinear characteristics compared to the crude oil price data.We then use the bestperforming benchmark model to predict the different modal components separately.Finally,the predictions of each modal component are summed to obtain the final prediction results.In this paper,we decompose the crude oil price series using the newly proposed median ensemble empirical modal decomposition(MEEMD)method,and then use the SVR algorithm to forecast each eigenmodal function and trend term from the decomposition,and finally integrate the results of each modal forecast.In this paper,a combined MEEMD＿SVR model is established to verify the effectiveness of the MEEMD decomposition algorithm in crude oil price forecasting,to compensate for the shortcomings of a single model,and to improve the accuracy of crude oil price forecasting.(2)Research on crude oil price prediction based on OPEC policy.The crude oil supply-demand relationship is a pivotal factor influencing oil prices,with supply-side factors largely determining the fluctuations in prices.Since 2016,the Organization of the Petroleum Exporting Countries(OPEC),the world’s largest oil-exporting organization,has coordinated and unified crude oil policies of its member countries through the OPEC+ meeting format,with the aim of more efficiently and reliably supplying crude oil to the global market and maintaining its stability.The oil supply of OPEC member countries holds significant sway over crude oil price fluctuations,however,quantifying the impact of these countries’ supply policies on oil prices remains a challenge,and little research exists on whether these policies can enhance crude oil price prediction accuracy.To address this gap,we propose a proxy indicator for OPEC crude oil policy and evaluate its efficacy in improving crude oil price prediction accuracy using multiple prediction models.Our empirical findings indicate that the crude oil price prediction models based on the decomposition algorithm and crude oil exporting country policy outperform the baseline model in terms of predictive performance.