| This thesis focuses on modeling natural gas price behaviour, and in particular mean reversion, seasonality, and the relationship between futures prices and spot prices. Since many of these features can not be adequately captured using only one source of randomness, as well as looking at 1-factor models, we also examine some 2-factor models and compare their effectiveness and explanatory power. By exploiting the relationship between futures prices and spot prices, we are able to recover information which can not be observed directly, and this enables us to perform maximum likelihood estimation for calibration of the models. |
