Oil Market Efficiency under a Machine Learning Perspective

Abstract

Forecasting commodities and especially oil prices have attracted significant research interest, often concluding that oil prices are not easy to forecast and implying an efficient market. In this paper, we revisit the efficient market hypothesis of the oil market, attempting to forecast the West Texas Intermediate oil prices under a machine learning framework. In doing so, we compile a dataset of 38 potential explanatory variables that are often used in the relevant literature. Next, through a selection process, we build forecasting models that use past oil prices, refined oil products and exchange rates as independent variables. Our empirical findings suggest that the Support Vector Machines (SVM) model coupled with the non-linear Radial Basis Function kernel outperforms the linear SVM and the traditional logistic regression (LOGIT) models. Moreover, we provide evidence that points to the rejection of even the weak form of efficiency in the oil market.

An Auxiliary Index for Reducing Brent Crude Investment Risk—Evaluating the Price Relationships between Brent Crude and Commodities

“…Today, oil prices still fluctuate greatly due to uncertain conditions and events. Athanasia et al (2019) provided evidence that points to the rejection of even a weak form of efficiency in the crude oil market, implying that a ceiling exists in predicting oil prices [3]. Therefore, it is necessary to surmount this ceiling in order to reduce crude oil investment risks by finding other auxiliary methods.…”

Robo-Advisors: Machine Learning in Trend-Following ETF Investments

“…value from trading data X(i), and K(X(i), X) represents the kernel function. As Dimitriadou et al [13] found that the SVM model combined with the non-linear Radial Basis Function (RBF) kernel exhibits better performances than any other kernels (e.g., linear, sigmoid), we use Gaussian Radial Basis Function…”

Money Neutrality, Monetary Aggregates and Machine Learning

“…We do not use log returns for the monetary aggregates because we would be testing for money In brief, the cross validation is performed to avoid overfitting of our forecasting methodology to the training dataset. The optimal parameters of our forecasting methodology are identified in a coarse-to-fine grid search to avoid exhaustive search, which is computationally enormous to handle [20]. The SVM methodology finds the optimal linear classificator by maximizing the distance of the marginal data points between the two classes; when the problem is not linearly separable, the kernelization of the system (the projection of the data space into a higher space using a kernel function) allow us to search for non-linear classifiers.…”