Seminarium i matematisk statistik
onsdagen den 31 maj 13.15 i MH:227

Slobodanka Mijalkovic
presenterar sitt examensarbete

Titel:
Modellering av finansiella instrument på den nordiska elmarknaden

Abstract:
In this Master Thesis the future spot price is to be modeled and
predicted using averaged daily prices collected between January1, 1996
and December 31, 1999. A few different time-series models, including
SARIMA7 (p,0,q), ARMAX (p,q) and ARIMA (p,1,q ), were applied to these
data. Three characteristic features, a seven day periodic cycle, level
changes and outliers, have been taken into account. Reasonable results
are found, however, signs of nonlinear behavior are found, implying that
a more complex model is needed in order to better fit data.

In the second part of the study, daily forward prices, collected between
October 28, 1997 and August 11, 1999, were used to find trends and
trends shifts in the forward prices. These discrete data include only
workdays, i.e. there is no data for weekends and holidays. For these
data, the double exponential smoothing method is rejected and there is
an evident need for more complex models. The possibility to use spot
price data for modeling forward data was also investigated. However,
also here a linear model is found not to be sufficient and more complex,
non-linear model should be used.