Predictive Modeling of Energy Derivatives Prices Using Machine Learning Techniques

This Master's thesis investigates the application of machine learning techniques in forecasting oil futures prices, with a particular focus on WTI crude oil contracts. The primary objective is to evaluate whether advanced models can outperform traditional time series benchmarks in short-term price prediction. The thesis begins with a comprehensive literature review covering oil derivatives pricing, machine learning modelling methods and commodity markets. The empirical part compares the performance of several forecasting ensemble and neural network models against benchmark time series model. Using historical daily price data, the models are assessed with standard error metrics to evaluate prediction accuracy. The results reveal that ensemble models consistently outperform traditional benchmarks. In contrast, LSTM models do not demonstrate superior performance over the benchmark.