EViews.com

Hi.

I am doing an analysis over imbalance prices in the electricity markets.
These prices are characterized by being quite stable most of the time, and then suddenly showing extreme deviations from the trend.

I am wondering which model / setup would be ideal for this analysis - anyone got any input?

Thanks a lot in advance!

Best Regards
J.E.R.

I should mention, that I have hourly data for the prices. Normally, the prices are the same in several hours, then jumping to another level for some time, then jumping back... and then, between all these jumps which occurs every few hours, extreme spikes are sometimes observed.

A typical day could look like this:

Hour 1: 26
Hour 2: 27
Hour 3: 27
Hour 4: 27
Hour 5: 77
Hour 6: 77
Hour 7: 77
Hour 8: 34
Hour 9: 34
Hour 10: 77
Hour 11: 77
Hour 12: 150
Hour 13: 77
Hour 14: 77
Hour 15: 77
Hour 16: 33
Hour 17: 33
Hour 18: 33
Hour 19: 33
Hour 20: 77
Hour 21: 33
Hour 22: 77
Hour 23: 33
Hour 24: 33

The spike-price in the example above, would be the 150-price seen in 1 hour. Sometimes these prices can get even higher - 300-ish.

I am interested in ways for trying to predict when these spikes happen, and have some data for explanatory variables I would like to include in my regression.
Any suggestions for which model I could use?

Maybe GARCH-models?

I think you should try to search for the so called jump-diffusion processes.
Perhaps this might be a good start: http://citeseerx.ist.psu.edu/viewdoc/do ... 1&type=pdf