I've discussed aspects of forecasting in several blog posts, including measures to reflect accuracy. Economists at the New York Fed have an insightful blog post on forecasting the lost recovery of the global financial crisis. They discuss the unprecedented monetary policy of the time, and with them, the challenges of forecasting using a state-of-the-art macro technique called dynamic stochastic general equilibrium modeling. This is not back-testing. This is forecasting (at the time) compared with how the economy actually evolved. It is well worth your time.