Search

Browse

• Communities
• Authors
• Year
• Titles
• Subjects

My Space

• Login

Errors in crude oil price forecasting on short-term basis ​

Al-Alami Bellal

Full record

---

Summary

This thesis forecasts short-term crude oil price changes, based on several time-series forecasting models. The crude oil prices which are used are monthly crude oil spot prices quoted in US Dollar of a benchmark consisting of Brent, Dubai and West Texas Intermediate "WTI" equally weighed. The numerical input data for forecasting are monthly crude oil prices of the mentioned benchmark for 30 year between [1988-2018].This thesis also addresses a careful qualitative and descriptive analysis of crude oil prices between 1988 and 2018, divided Into 8 periods, each is decided based on significant events that occurred within each period that influence the analysis leading to observable fluctuations in prices.In this thesis I have used four different approaches for forecasting process of monthly crude oil prices including conventional forecasting approach, inflation-adjusted prices approach, unconventional "month by month" forecasting approach and the unconventional hybrid forecasting approach using the moving average and exponential smoothing models. The accuracy of all generated models was evaluated using different error measures including "MAD, MSE and MAPE", to investigate the capability of generated forecasting model in accurately forecasting the monthly crude spot prices.Afterwards, a numerical comparison of the unconventional hybrid approach and the month by month approach to the conventional and inflation-adjusted approach is made. The error estimates of the different forecasting models show that unconventional approaches tend to provide significant improvement in forecasting results and can generate larger reductions in error values than those generated by conventional and/or the inflation-adjusted approaches.My understanding for forecasting of crude oil prices on short term basis is that using direct numerical forecasting methods solely will generate models and forecasted results that suffer large errors deviating than those real ones, the sensitivity of oil industry and oil prices to external events and factors make the process of forecasting uneasy, and therefore a researcher must find ways that both can combine the numerical methods but with twists and may be combine different approaches to enhance these results. Additionally, a qualitative judgment of experts can play a good role while forecasting by watching and monitoring the global trend, regional events, political situations, environmental aspects, presence of technological advancements, seasonality and consumer behavior, then take into account all these factors while forecasting specially that oil prices are heavily impacted by global oil supply and demand.

Available Files

Services