Siting offshore wind farms is a complex problem due to the wake interactions between wind farms. We develop a profit maximizing portfolio model based on underlying network models to track the wake effects through a series of wind farms. Our portfolio model optimizes the siting of wind farms considering multiple wind directions and wind speeds and performs better than simple decision heuristics. Excluding sites from the portfolio has nonlinear impacts on the profitability of the portfolio of sites in that areas excluded from consideration have greater impacts on profit if they are grouped together or aligned parallel to the prevailing wind direction. The model can be readily adapted to include additional cost factors.