Need some input in how to attack this problem. Given are 8 timeseries:
- UK Oil price, Delivery Quarter 1 2020
- UK Oil price, Delivery Quarter 2 2020
- UK Oil price, Delivery Quarter 3 2020
- UK Oil price, Delivery Quarter 4 2020
- FR Oil price, Delivery Quarter 1 2020
- FR Oil price, Delivery Quarter 2 2020
- FR Oil price, Delivery Quarter 3 2020
- FR Oil price, Delivery Quarter 4 2020
The prices plotted (UK=BQ)
The question now is how to price the right to transfer oil from UK to FR. For the year 2020. Context: If UK price is above FR price, assume we can buy at the spot in FR and sell at the UK spot. If we have this transfer right.
The transfer right is bought for a full year and you pay whether you use it or not. You buy a daily transfercapacity and you can choose to not use it (if UK is cheaper than FR for example).
Transfer is only one way. Delivery can be executed between the first day of the quarter and the last (Q1 starts Jan 1).
I plotted the UK-FR spreads:
Now my question is whether there is a good way to attack this problem.
I think that it is smarter to model the UK-FR difference than the underlying timeseries itself. Is that correct?
Also when it comes to modeling this, my intuition says 390 days is not a lot. In previous problems I had multiple years of data so I could better detect seasonality and trends with a STL decomposition. Here that does not seem to make sense.
Any input or ideas are welcome.