Extracting Knowledge from Technical Reports for the Valuation of West Texas Intermediate Crude Oil Futures

Abstract

This paper proposes and demonstrates an approach for the often-attempted problem of market prediction, framed as classification task. We restrict our study to a widely purchased and well recognized commodity, West Texas Intermediate crude oil, which experiences significant volatility. For this purpose, nine learners using features extracted from monthly International Energy Agency (IEA) reports to predict undervalued, overvalued, and accurate valuation of the oil futures between 2003 and 2015. The often touted “Efficient Market Hypothesis” (EMH) suggests that it is impossible for individual investors to “beat the market” as market and external forces, such as geopolitical crises and natural disasters, are nearly impossible to predict. However, four algorithms were statistically better at the 95% confidence interval than “Zero-Rule” and “Random-Guess” strategies which are expected to pseudo-reflect the EMH. Furthermore, the addition of text features can significantly improve performance compared to only using price history from the oil futures data, challenging the validity of the semi-strong versions of the EMH in the crude oil market.