KTH Matematik  

Matematisk Statistik

Tid: 26 augusti 2015 kl 16.15-17.00.

Seminarierummet 3721, Institutionen för matematik, KTH, Lindstedtsvägen 25, plan 7. Karta!

Föredragshållare: Susanna Kaas

Titel: Validation of market commodity forward curves (Master's thesis)

Abstract In this thesis the aim was to propose a method that could be used to validate the market commodity forward curve and analyse if the method is possible to apply. The thesis is limited to forward curves with equally spaced maturities up to one year and seasonal price patterns. The method proposed was to simulate a reference curve using theoretical methods and compare the reference curve to the corresponding market curve. It is suggested to construct the reference curve by simulating futures price with the seasonal cost-of-carry model and perform linear interpolation between these simulated values. The validation method was chosen to apply on futures written on UK natural gas traded on Intercontinental Exchange for every trading day in December 2013. The historical data set was based on settlement prices for the period 2011-01-01 to 2013-11-30. The resulting reference curves seemed to capture the seasonal behaviour of UK natural gas in a correct way and the shape of the curve seemed to follow the market curve. However results showed that time series generated from historical data representing the state variables did not fulfil model assumptions for all cases. Therefore the observation period was shortened to 2012-07-01 to 2013-11-30 but the result was only slightly improved. It was still the case that some of the state variable processes did not follow model assumptions. By performing likelihood ratio test it was found that for some state variables the speed of mean reversion could be set to zero. The conclusion was drawn that the proposed method is not appropriate to use for validating the market curve for the considered contracts. This is because model assumptions for state variables were not always fulfilled and some of the state variable process could be reduced to random walks. It could be the case that model assumptions are fulfilled if the method is applied to another month. However it is difficult to use a method for validation if historical data sometimes suggest that times series are not stationary and do not fulfil model assumptions. Hence for the chosen commodity the validation method does not seem applicable.

The full report (pdf)

Till seminarielistan
To the list of seminars

Sidansvarig: Filip Lindskog
Uppdaterad: 25/02-2009