dr = pfolio_var - (weights[0] ** 2 * PG_var_a) - (weights[1] ** 2 * BEI_var_a) Why do you have to square the weights?

Hi Marcus! Thanks for reaching out! Because what we actually need to do is take the square of the standard deviation times the respective weight value. Since we are using the value of the variance directly (which equals standard deviation squared), what is left for us to do is take the square of the weights as well. Hope this helps. Best, Martin