I would like to generalize Paul Teetor's A Better Hedge Ratio, which uses prcomp() to determine the orthogonal regression ratio between two variables. I am hoping to generalize this to multiple variables, but am having trouble finding anything in the literature. Any suggestions?

  • $\begingroup$ This question is not clear at all. What is TLS (yes, I can and in fact did look in the linked paper, but the question should be clear on its own)? How do you want to extend it to multiple variables? What do you mean by having trouble "finding" accurate results? In the literature? $\endgroup$
    – amoeba
    Dec 29, 2014 at 12:45
  • $\begingroup$ Thanks for the edits, but I am still not sure what exactly you have in mind when you say "generalize this to multiple variables". What would be hedge ratio for multiple variables? Given that "ratio" is usually understood between only two. Are you thinking of multiple regression (so looking for something that could be called "multiple orthogonal regression")? $\endgroup$
    – amoeba
    Dec 29, 2014 at 14:26
  • $\begingroup$ The r-code in the paper is for two variables. It is followed by saying that this same procedure can be generalized for multiple variables, but I am not sure how to write the code. $\endgroup$
    – Stu
    Dec 29, 2014 at 16:09
  • $\begingroup$ Stu, I have found a solution. I provided an extensive answer in another thread, and I vote to close this one as a duplicate. Please see my answer there, I hope it will answer your question. $\endgroup$
    – amoeba
    Feb 7, 2015 at 0:58


Browse other questions tagged or ask your own question.