This study examines the accuracy of five competing naive and futures-based localized cash price forecasts. The third week's price for each month from 1987-1996 is forecasted from vantage points one to 11 months preceding the observed price. Commodities examined included those relevant to Kansas producers: the major grains, slaughter steers and slaughter hogs, several classes of feeder cattle, cull cows, and sows. Information about relative forecasting accuracy across forecast methods was collapsed into regression models of forecast error. The traditional forecast method of deferred futures plus historical basis had the greatest accuracy -- even for commodities that are substantially different from those specified in related futures contracts. Adding complexity to forecasts, such as including regression models to capture nonlinear bases or biases in futures markets, did not improve accuracy.