Predicting future Brent oil price on global markets

Marek Vochozka, Jakub Horák, Tomáš Krulický, Pedro Pardal

Predicting future Brent oil price on global markets

Číslo: 3/2020
Periodikum: Acta Montanistica Slovaca
DOI: 10.46544/AMS.v25i3.10

Klíčová slova: Artificial neural networks, time series, commodity, prediction of future development, global market, experiment

Pro získání musíte mít účet v Citace PRO.

Přečíst po přihlášení

Anotace: International oil prices affect the development of many areas of the

world economy, where most predictions of macroeconomic
variables depend on the changes in oil prices. The high volatility of
oil prices is a cause of prediction complexity, especially in
crisis or during the current coronavirus pandemic. In this paper, a
special and promising type of artificial neural network
(Long Short-Term Memory) is used for predicting oil prices. The
paper's objective is to predict the development of B
daily values to 30 June 2021. For this purpose, available data for the
period from the end of June 1988 to the end of November 2020 was
used. To achieve the objective of the paper, two research questions
were formulated: whether the created specific neural network is a
suitable tool to smooth the time series of Brent oil prices, i.e., a
suitable tool to predict the future development of the price of this
commodity, and what development of oil price can be expected with
regard to the current situation in global markets. It was confirmed
that each of the networks retained could smooth the time series
successfully, and it can make a reasonable prediction of the future
Brent oil price development. This paper's primary finding is that the
created neural network with integrated LSTM can be used for
predicting Brent oil prices. As for the further development of oil
price, the oil market will also respond to the positive development
of the economy and the growth of the world economy's overall
product, where both the quantity of extracted oil in the market and
its price will grow.