What does the (Expected return - 5%)/std dev has anything to do with the suggested solution? Initially, I thought they assume it's CAPM so I used the formula for the proportion of A as: x_a = (V_b - Covariance)/(V_a + V_b - 2Covariance) such that x_a + x_b = 1 but obviously I got tricked.
S2006 Q6 ii (E - 5)/ sigma is called the Market price of risk and you are dead right that we use this in CAPM. You will see it in the capital market line on page 43 of the Tables. MVPT, and finding the efficient frontier, is about maximising expected return for a given level of risk or, equivalently, minimising risk for a given level of expected return. Whatever way you look at it, we can achieve this by maximising the market price of risk. So, we express MPR as a function of xA and then choose xA to maximise this fucntion. ie differentiate set to zero etc. John