Modelo de previsão de preço futuro de petróleo bruto nos Estados Unidos
Carregando...
Autores
Orientador
Bortoluzzo, Adriana Bruscato
Co-orientadores
Citações na Scopus
Tipo de documento
Data
2023
Título da Revista
ISSN da Revista
Título do Volume
Resumo
A indústria do petróleo ocupa um lugar de importância na economia mundial;
segundo a Statistical Review of World Energy, publicada em 2021, mais de 80% da
energia consumida no mundo é fruto de combustível fóssil. Dentro da indústria do
petróleo, a economia americana e, consequentemente, o dólar ocupam um lugar de
importância na indústria do petróleo principalmente após o ano de 1974. O
entendimento do comportamento do petróleo e como ele se relaciona com o mercado
financeiro é importante não apenas para empresas produtoras de mercadorias ou
commodities, que são afetadas diretamente pela variação dos preços, mas também
para investidores da bolsa, gerentes de portifólio e especuladores em geral; o petróleo
é chave para a construção de portifólios bem diversificados e para decisões
financeiras que maximizam retornos e otimizam as decisões de risco. O principal
objetivo deste trabalho é examinar diferentes modelos de previsão para os preços
futuros do petróleo nos Estados Unidos. Três modelos são examinados para identificar
empiricamente o modelo com maior precisão e melhores resultados preditivos. Foram
feitas previsões sob os modelos Autorregressivo e de Médias Móveis Integrado
(ARIMA), Vetor Autorregressivo (VAR) e de Random Forest para os períodos de 01
de janeiro de 2019 a 01 de dezembro de 2019 e 01 de julho de 2021 a 01 de junho de
2022. O modelo VAR, se comparado aos três modelos selecionados e com os
parâmetros utilizados, conseguiu descrever melhor o cenário econômico e o petróleo
como um agente econômico e financeiro, levando em consideração seus
componentes e relação com outros fatores. Contudo, o modelo de Random Forest tem
maior sensibilidade em capturar as movimentações e os choques no preço,
demonstrando quedas e altas ao longo do tempo.
The oil industry occupies an important place in the world economy; according to the Statistical Review of World Energy, published in 2021, more than 80% of the energy consumed in the world comes from fossil fuel. Within the oil industry, the American economy and, consequently, the dollar take over a place of importance in the oil industry, mainly after 1974. Understanding the manners of oil and how it relates to the financial market is important not only for companies that produce goods or commodities, which are directly affected by price fluctuations, but also for stock market investors, portfolio managers and speculators in general; Oil is key to building well diversified portfolios and making financial decisions that maximize returns and optimize risk decisions. The main objective of this work is to examine different forecast models for future oil prices in the United States. A variety of models are examined to empirically identify the model with the highest accuracy and best predictive results. The forecasts were constructed using the Integrated Moving Averages (ARIMA), Vector Autoregressive (VAR) and Random Forest models for the time spans from January 1, 2019, to December 1, 2019, and July 1, 2021, to June 1 from 2022. The VAR model, compared to the three selected models and with the parameters used, was able to better describe the economic scenario and oil as an economic and financial agent, considering its components and correlation with other factors. On the other hand, the Random Forest model is more sensitive in capturing movements and price shocks, demonstrating declines and rises over time.
The oil industry occupies an important place in the world economy; according to the Statistical Review of World Energy, published in 2021, more than 80% of the energy consumed in the world comes from fossil fuel. Within the oil industry, the American economy and, consequently, the dollar take over a place of importance in the oil industry, mainly after 1974. Understanding the manners of oil and how it relates to the financial market is important not only for companies that produce goods or commodities, which are directly affected by price fluctuations, but also for stock market investors, portfolio managers and speculators in general; Oil is key to building well diversified portfolios and making financial decisions that maximize returns and optimize risk decisions. The main objective of this work is to examine different forecast models for future oil prices in the United States. A variety of models are examined to empirically identify the model with the highest accuracy and best predictive results. The forecasts were constructed using the Integrated Moving Averages (ARIMA), Vector Autoregressive (VAR) and Random Forest models for the time spans from January 1, 2019, to December 1, 2019, and July 1, 2021, to June 1 from 2022. The VAR model, compared to the three selected models and with the parameters used, was able to better describe the economic scenario and oil as an economic and financial agent, considering its components and correlation with other factors. On the other hand, the Random Forest model is more sensitive in capturing movements and price shocks, demonstrating declines and rises over time.
Palavras-chave
Titulo de periódico
URL da fonte
Título de Livro
URL na Scopus
Idioma
pt
URL permanente
Notas
Membros da banca
Bortoluzzo, Adriana Bruscato
Artes, Rinaldo
Bortoluzzo, Maurício Mesquita
Área do Conhecimento CNPQ
CIENCIAS SOCIAIS APLICADAS