Could somebody tell me whether suggestions in bold true or not?
Chapter 2.2 Interpretation of the eigenvectors/eigenportfolios
This paper says that loadings in the maximal eigenvector need to be all positive and should not change sign, what if i have negative ones, can i force them to be positive always by simply taking them by module, e.g. MathAbs(Vector) ?
Q # 2 : The same paper also defines weights for eigenportfolio in this way :
Q[i] = EigVecCoef[i] / StdDev[i] // Page # 10 paragraph 2.2 in the doc above
There is also another paper that says that eigenvector is an angle (or direction) of the portfolio's spread which allows to map current portfolio's spread to initial axes (dimensions) :
So i do not understand - why do i need to divide each value in eigenvector by standard deviation to calculate weights if this portfolio is already mapped to initial axes?