Issue with computation of the H matrix in msfc.R code

[edoscm]

First of all, thank you for the great package, I found it very useful for my thesis!

I was reading the Benth's article "Extracting and Applying Smooth Forward..." and I noticed that, in the computation of the matrix H, h_j's coefficients are, for example, h_j[1][1] = 28,8*(k[j+1]**5 - k[j]*5) and not 28,8(k[j+1] -k[j])**5 .

So, I would ask you if it is possible or is it my misunderstanding.

[sleire]

Hi @edoscm, and thank you for your interest in the etrm package. I have created a new branch for evaluating the computation of the H matrix. Feel free to explore it. I will look more into this when I have available time (in the final phase of my phd right now).

FYI: I have written a paper titled "etrm: Energy Trading and Risk Management in R". It will be published in the next issue of the R-journal (June/ July 2022). This will describe the package in further detail. It also presents some ideas for further work, if anyone would be interested in participating in the project.

Again, thank you for your interest, I will look into the calculation of the H matrix.

[Jidvei]

First of all, thanks for your great work, and best of luck with the finalization of your thesis!

What edoscm points out is correct, and the change in the branch should be correct.
However, your default example does no longer compute, due to what I'd assume is numbers assume extreme values(?).

Moreover, the solutions I obtain doesn't seem to be exact (i.e. instead of fwdprice 400 then it finds 400.48 etc). Is this by design?

[sleire]

Hi, and thank you for your interest in etrm!

The computed prices should be close to/identical to market prices. I believe further modifications to https://github.com/sleire/etrm/blob/issue-4-msfc-hmatrix/R/msfc.R are needed in order to implement the adjusted calculation described above. I will explore this when I get more time, but this may take a while. Suggestions for changes and pull requests would be highly appreciated.

There are also several new features we would like to add to etrm, these are described in the R-journal paper (see link in Readme).