Font Size: a A A

Prediction of short-term natural gas prices using econometric and neural network models

Posted on:1999-01-23Degree:Ph.DType:Dissertation
University:Texas A&M UniversityCandidate:Reiter, Doris FFull Text:PDF
GTID:1462390014971822Subject:Economics
Abstract/Summary:
During the last decade, the natural gas industry underwent a thorough deregulation process after being subjected to federal oversight for more than 40 years. Federal deregulation provided the basis for a new market, which is now more than ever before being driven by competition in the production, transportation and distribution sectors. Now, lack of price transparency causes high fluctuations in the spot market, which exposes market participants to a considerable price risk. Two strategies are necessary for successful price risk management: fundamental analysis, which attempts to understand the market principles and interactions; and quantitative analysis, which develops models to predict price developments using the fundamental analysis as a basis. A need for short-term solutions is now obvious.; Therefore, this work, on the one hand, gives a comprehensive description of the current state and market operations of the natural gas industry. On the other hand, the understanding of the natural gas industry obtained in the fundamental analysis provided the background for the development of two models for day-to-day price prediction. An econometric and a neural network model were chosen as the best solutions. For practical application purposes, the prediction performance of the models was tested in a simulated trading scenario and compared to a best and worst case scenario. We found that both models created profits during the time of the test. The neural network model showed better results than the econometric model. Also, in comparison to a scenario with perfect prediction quality and a "guessing" scenario, the results were very pleasing. These models can provide a valuable, supportive tool for trading in a speculative environment.
Keywords/Search Tags:Natural gas, Models, Neural network, Price, Prediction, Econometric, Scenario
Related items