This dissertation investigates the optimal investment region for a pension fund under two different economic models: the Wilkie model and a correlated geometric Brownian motion model. Using a genetic algorithm and dynamic programming, the author searches a restricted space of investment strategies to determine the optimal region that minimizes downside risk, as measured by the median lifetime and quantiles of the pension fund value. The results show that while the optimal regions and quantiles differ slightly between the two models, both models identify a similar optimal region of investment over the lifetime. Splitting the investment period into accumulation and decumulation phases is also briefly examined. The dissertation concludes that the simpler geometric Brownian motion model can provide similar insights to the more complex Wilkie model,