A Queueing Theory Approach to Forecasting Oil Futures Prices

Abstract

Since the 1990s, oil futures have become a part of an increasing number of investors’portfolios. Whereas in the past, the prices of oil futures only reflected the valuationof oil, futures contracts have become a means through which investors speculate onthe future prices of oil. As a result, oil futures prices have begun to take on thecharacteristics of financial assets, such as stocks and bonds. In this thesis, we createa model to forecast the price of oil futures. Our model takes into account the changingcharacteristics of oil futures by using a market microstructure approach to examineprice movements. We do this by developing a strategy to model the arrival rates ofbuy and sell orders, and then to model the impact of those arrivals on the price of thefutures contract. We find that although the models in most literature approximateorder arrivals as Poisson processes, trades arrive in clusters, making such a modelinaccurate. We also find that a linear regression is rigorous enough to capture therelationship between trade imbalance and price movement.