 Exponential-utility trading with Gaussian increments

With a unit cash asset, independent Gaussian price increments of mean $\mu$ and positive variance ${\sigma }^2$, and exponential terminal utility $-e^{-\gamma x}$, the optimal predictable risky holding is the displayed constant number of shares. The one-step exponential-loss exponent is $-\gamma \mu \pi +{\gamma }^2{\sigma }^2{\pi }^2/2$; completing its square gives a minimal multiplier $e^{-{\mu }^2/(2{\sigma }^2)}$. Backward induction proves optimality among adapted strategies and gives value $-e^{-\gamma x-T{\mu }^2/(2{\sigma }^2)}$.