Suppose a risk-averse investor with differentiable Bernoulli utility $u$ and wealth $w$ that can be allocated between asset $X$ and $Y$. Both $X,Y$ have positive expectations. If the investor invests a fraction $a$ of their wealth in $X$ and $1-a$ in $Y$, their wealth will be $aXw+(1-a)Yw$ in the next period.
How do I argue that $a=0.5$ is the optimal? I think I need to compare the random variable $Z_a=aX+(1-a)Y$ to $Z_{0.5}=0.5X+0.5Y$. How do I proceed?