Machine Learning Heating Oil Price Forecasts

Source abstract

Historically speaking, a significant number of market players have put a great deal of significance on the capacity to reliably estimate the prices of energy commodities. In order to discover a solution to the issue, our inquiry looks into the monthly price of heating oil as it applies to New York Harbor No. 2. The price series under investigation has major economic repercussions, and the sample under review runs nearly 40 years, from June 1986 to March 2024. In this scenario, cross-validation methods and Bayesian optimization methodologies are applied to develop Gaussian process regression models, which are used to produce price estimates. For the out-of-sample testing period of September 2016–March 2024, our empirical prediction technique produces fairly accurate pricing projections, as indicated by the relative root mean square error of 1.4817%. Price prediction models give governments and investors with the knowledge they need to make smart judgements regarding the heating oil market.