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]
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?