Two risky assets have return rates r1, r2 that are random variables with expected values r1 and r2 respectively. Let the matrix of covariances be: [sigma1,1 sigma1,2; sigma2,1 sigma2,2]. Because sigma1,2 = E[(r1 - r1)(r2 - r2)] and sigma2,1 = E[(r2 - r2)(r1 - r1)], (1) sigma2,1 = sigma1,2. Consider the portfolio composed of these two assets with weights w1, w2. Then the variance of the portfolio is (2) sigmaP^2 = sum(i=1 to 2) sum(j=1 to 2) sigma i,j wi wj. The expected return rate for the portfolio is (3) rP = w1r1 + w2r2. Let rf be the return rate for a risk-free asset. The portfolio F of the one fund theorem, is the combination of risky assets that maximizes (4) (rP - rf) / sigmaP subject to the constraint