Oil market modelling: A comparative analysis of fundamental and latent factor approaches

Abstract

We formally compare fundamental factor and latent factor approaches to oil price modelling. Fundamental modelling has a long history in seeking to understand oil price movements, while latent factor modelling has a more recent and limited history, but has gained popularity in other financial markets. The two approaches, though competing, have not formally been compared as to effectiveness. For a range of short-medium-and long-dated WTI oil futures we test a recently proposed five-factor fundamental model and a Principal Component Analysis latent factor model. Our findings demonstrate that there is no discernible difference between the two techniques in a dynamic setting. We conclude that this infers some advantages in adopting the latent factor approach due to the difficulty in determining a well specified fundamental model.

Uncovering predictability in the evolution of the WTI oil futures curve

“…Based on this literature we adopt the fundamental commodity futures forecasting framework of Andreasson et al (2016), as well as the principal component approach of Chantziara and Skiadopoulos (2008) as two benchmarks for our proposed FTS model. This aligns with the approach taken by Cummins et al (2016) who adopt both frameworks and find no discernible difference between the performance of the latent and fundamental models.…”

“…To alleviate any such concerns, we concentrate only on the most actively traded generic monthly futures in line with Cummins et al (2016). The futures we use are CL1-CL9, CL12 and CL18 obtained from the CME Group.…”

“…where P 1 = 99%, 1 { λ k >0} is to exclude possible zero eigenvalues, and 1{·} represents the binary indicator function. When we align with Chantziara and Skiadopoulos (2008) and Cummins et al (2016) and utilise three principal components, 99% of the variation in our dataset is explained; therefore we also specify P 1 = 99%.…”

“…First, we outline the number of principal components used. In line with Chantziara and Skiadopoulos (2008) and Cummins et al (2016) we specify three discrete principal components to model the oil futures curve. This results in 99% of the variance in the data being explained.…”

“…Secondly, it facilitates the modelling of data observed at a small number of unequally spaced points, as well as those with possible non-negligible measurement error, which are common in business statistics (see, e.g., Kowal et al 2018). In our specific context, we follow Cummins et al (2016) in focusing on the most actively traded futures contract expiries (CL1-CL9, CL12, and CL18), resulting in unequally spaced data points along the futures curve, which our FTS framework models effectively.…”